3.1 Acting as if

What we might call the ‘as if’ objection rejects (B) and affirms:Footnote ¹¹

As such, the objection suggests that we should understand Belief as acting as if the Highest Good is attainable.Footnote ¹² In willing the Highest Good, it is rational for us to act as if it is attainable. In so doing, we are rational because as long as it is valuable for us to act as if we can completely attain the Highest Good, we are warranted in doing so. For instance, even though I might know that it is impossible to publish a book that is completely error-free, it is valuable for me to act as if I can do so: I can always work to make the book better – by meticulous editing, fact-checking, and so on – which is, overall, a valuable project.

A few passages even support reading Kant as committed to this view. In the Progress essay, Kant states that, for practical reasons, we ought to ‘act, as if we knew that these objects [God and the immoral soul] were real’ (Prog, 20: 298). In the Jäsche Logic, he claims that the proof of God’s existence has only ‘a practical purpose, i.e., to act as if there is a God’ (JL, 9: 93).

However, an obvious difficulty with the as if objection is that Kant’s moral proof aims to deliver rational Beliefs concerning the attainability of the Highest Good and the existence of God, which the as if objection does not preclude. As we saw, Kantian Belief is a mode of assent. Assents are propositional attitudes, not acts. As such, the as if reading is at odds with Kant’s explicit statements that Belief is an assertoric propositional attitude, not ‘acting as if’ (see JL 9: 66; KpV, 5: 134). The issue is not merely textual. Acting – as well as acting as if something is true – is compatible with holding certain propositional attitudes. Thus, we can ask, even if we act as if the Highest Good is attainable, what propositional attitudes should we hold? So, even if (B_{as if}) is true, we can ask whether (B) is true.

Given that the as if objection does not count for or against Attainability’s plausibility, I will set it aside.

3.2 Modality

Modality objections reject (D) and instead affirm:

Accordingly, the objector claims that (B) is false, arguing that the rational pursuit of an end requires only that the subject assent to the logical possibility of attaining it – contrary to Kant’s view that the relevant modal notion is a practical possibility (i.e. that it is possible for us to attain the end through our actions). Thus, (B) should be replaced with:

For instance, Guyer claims that, for our pursuit of the Highest Good to be rational, all we need to know is that it is possible to achieve it, a condition that is ‘satisfied as long as [the Highest Good] is not self-contradictory’ (Guyer Reference Guyer2000: 361).Footnote ¹³ Thus, we would be rational to will the Highest Good as our end while assenting that it is merely logically possible to attain it.

The problem with the modality objection is that it conflates two kinds of ought propositions:

An example of a general ought-proposition is (1) ‘It ought to be that there are no famines’. The ‘ought’ is general: it does not specify any agent. For proposition (1) to be true, a state of affairs without famine need be merely logically possible. As such, (1) does not imply that a world without famine is physically possible, nor that we can do anything to bring that world about.

An example of an agent-relative ought-proposition is (2) ‘Biden ought to have prevented World War II’. The ‘ought’ is agent-relative: it is indexed to a particular agent at a particular time. For the proposition in (2) to be true, it needs to be the case that the agent to which it is indexed, at the relevant time, can bring about the relevant state of affairs. Thus, (2) is false. Biden could not have prevented World War II – and thus was not under an obligation to. Preventing a war in the past is not something Biden could have done. This is despite the fact that, presumably, it ought to have been the case that World War II never happened. The crux of the issue is not whether Biden preventing World War II is logically, and even metaphysically, possible. Rather, it is whether it is possible for Biden to bring about the relevant state of affairs.

Proponents of the modality objection are correct to point out that some ought-propositions require only logical possibility. Thus, we can express the Highest Good as a general ought-proposition. If that proposition – ‘the Highest Good ought to be attained’ – is true, then there must be a world that is at least logically possible in which the proposition that ‘the Highest Good is attained’ is true.

However, the modality objection mistakenly construes our obligation to will the Highest Good as a general ought-proposition that requires mere logical possibility, rather than an agent-relative ought-proposition requiring practical possibility. Of course, it is not solely in our power, or the power of any individual, to achieve the Highest Good. But this is precisely why, Kand holds that, we are rational to Believe in its attainability.

In sum, given that our rational obligation to pursue the Highest Good is an agent-relative ought-proposition, that obligation is constrained by practical possibility. Therefore, modality objections have little force against Attainability.

3.3 Approximation

A deeper source of worry comes from the ‘approximation’ objection, which rejects (A) and instead endorses:Footnote ¹⁴

On this basis, the approximation objector can reject (B) outright.

The objector argues that Kant’s moral argument makes it no easier to see why we should make the full realization of the Highest Good our end, rather than some more modest goal, such as, asymptotically approaching the Highest Good. The thought is that, given that ought implies can, any plausible moral theory can demand only what we are capable of doing. As a result, our moral obligation is to approximate the ideal of the Highest Good as best we can, not fully to realise it – given our finite capacity to affect the world. Thus, Belief in the attainability of the Highest Good is unnecessary because we can rationally approximate an end, despite knowing or believing (in the contemporary sense) that its full attainment is impossible. Returning to our book example, I seem rational in attempting to approximate writing a perfectly error-free book, while knowing that such perfection is unattainable.

One response to approximation objections argues that it is not always the case that we ought to approximate an end that is unattainable.Footnote ¹⁵ Suppose you are about to undergo surgery and require three doses of anaesthetic. However, the hospital is running low on the medication. In this case, the anaesthesiologist would be mistaken to assume that giving you two doses is better than giving you one or none. In the absence of the unattainable third dose, two doses might be useless (they may not put you to sleep) or even dangerous (you might wake up mid-surgery).

While I think responses along these lines are promising, they face an obvious difficulty. Even if it is true that we should not always approximate an end, this alone does not show that we should not do so in the case of the Highest Good. Thus, the approximation objection remains a powerful challenge to Kant’s moral argument and the necessity of Belief.

3.4 Not-believing objection

A further objection targets the epistemic attitude involved in (B), particularly the possibility of withholding belief about Attainability. What we might call the not-believing objection holds that (B) should be replaced with:

The objector maintains that (B) is false because it is rationally consistent to pursue an end so long as one does not believe that the end is impossible. According to this objection, Attainability is too demanding, as it requires a positive attitude towards the attainability of our ends, rather than merely the absence of a negative attitude.

Willaschek describes a case of non-believing:

Willaschek argues that Test Taker is a case in which a rational agent is ‘pursuing an end that [he] does not believe he can realize’ (ibid.). He distinguishes between merely doing something and trying, on the grounds that doing presupposes foreknowledge of success, whereas trying excludes such knowledge (p. 238). According to Willaschek, cases like Test Taker show that as long as one does not know that success is impossible, one can rationally refrain from ‘taking a stand at all’ on the possibility of success and still rationally ‘try’ to pass the test (ibid.).

I doubt, however, that the objection rests on an analysis of ‘trying’ for two reasons.Footnote ¹⁶ First, ‘trying’ is not Kant’s terminology. Willaschek suggests that trying can always be rational given the ‘importance of the end pursued and … the costs involved in trying’ as long as the chance of success is greater than zero and the ‘stakes are high enough’ (Reference Willaschek and Höwing2016: 239). This mischaracterises Kant’s moral argument. As we saw, Kant’s argument begins with the independent premise that we are rational to will the Highest Good because rational morality demands we ought to will the Highest Good – irrespective of cost.

Second, Willaschek’s use of ‘trying’ appears anachronistic. On a widely held view trying is, pace Willaschek, consistent with knowledge of success. Action theorists point out that knowing that I will succeed in raising my arm does not prevent me from trying to do so.Footnote ¹⁷ Moreover, it seems plausible that trying is also consistent with full-blown knowledge that success is impossible – again, pace Willaschek. For instance, one might try to push over a wall as part of an exercise programme, fully knowing that one lacks the strength to do so.Footnote ¹⁸

The core of the not-believing objection, then, rests on the claim that a ‘norm of consistency’ typical of structural rationality is not violated in cases like Test Taker (Willaschek Reference Willaschek and Höwing2016: 224). Structural rationality governs the coherence of our mental states. According to one widely accepted requirement: if an agent believes p and believes that p entails q, then the agent ought to believe q – provided their beliefs are transparent to them.

Kant writes that, in willing an end, ‘the use of means is already thought, and the imperative extracts the concept of actions necessary to this end’ (G, 4: 417; my italics). Interpreted in terms of contemporary belief, Kant is affirming an epistemic claim: if an agent wills an end, they believe that the means, M, is necessary for the end, E (see Lee Reference Lee2018). This belief does not logically commit the agent to believing that E is possible through their actions, but only that if E is to occur, it must (at least partially) involve M – namely, their actions. As such, if a subject wills an end, at most this commits them to withholding judgement about E’s attainability. According to the objector, then, what is clearly not at issue is any failure of structural rationality.

In sum, the approximation and not-believing objections present potentially powerful critiques of Attainability, and thus of Kant’s moral argument in the Critique of Practical Reason, whereas the ‘as if’ and modality objections pose less substantial concerns.

4.1 Practical considerations

The first element of my proposal concerns the practical consideration underlying Attainability, and thus the kind of constraint it involves. The most direct evidence of Kant’s commitment to Attainability appears in the section ‘On Assent from a Need of Pure Reason’ in the Critique of Practical Reason. There, Kant offers the following as the practical consideration motivating our assents to Attainability:

The first clue to the kind of constraint at issue is that the passage makes no reference to rational consistency. Kant holds that agents cannot make something the object of their will unless they assent that it is possible – or more precisely, practically possible (i.e. attainable through their actions). Such considerations are, as he notes, ‘quite independent’ of an agent’s duty to will the Highest Good (KpV, 5: 142). Thus, even if one does not assent to Attainability, one would still be obligated to will the Highest Good.

Kant frequently states that the practical consideration which rationalizes Belief is a ‘practical relation’ between our assents and the necessary means to our ends (A823/B851).Footnote ¹⁹ He defines this relation in terms of a means–end structure: ‘[o]nce an end is proposed, then the conditions for attaining it are hypothetically necessary’ (ibid.). In the Critique of Practical Reason, he maintains that certain assumptions serve as the ‘means of promoting what is objectively (practically) necessary’ (KpV, 5: 146).Footnote ²⁰ Similarly, in the Jäsche Logic, Kant is reported as saying: ‘I see myself necessitated [genöthigt] through my end … to accept [anzunehmen] as possible a highest good in the world’ (JL 9:67n).

Consider a case of action: if I set myself the end of going to the city, a necessary means to pursuing that end is transporting myself there. The relation between my end and the means of achieving it is practical – my act of transportation (e.g. catching a train, riding a bike) enables me to pursue the end. Because catching the train (or some other form of transport) is necessary for pursuing that end, it is a practically justified action. In a broad sense, as the passages quoted above suggest, Kant holds that the means necessary for pursuing ends include propositional attitudes as well as actions. Thus, he conceives of means–end relations involving practically justified assents in much the same way as those involving actions: assent can function as a means to achieving an end, and the relation between ends and means can serve as justificatory grounds for that assent.

Kant, however, cautions against assuming that practically justified assents are necessary for action as such: ‘If we look merely to actions, we do not need this Belief’ (JL, 9: 68n). This warning reflects the plausible view that we do not treat propositional attitudes as means to actions. After all, it is unclear how a belief could serve as a means to something like raising my arm.Footnote ²¹

This warning, however, does not extend to willing:

Here, Kant again insists that the practical consideration in light of which we assent to Attainability is that such assents are necessary for willing. Thus, the most natural way to interpret the relevant practical consideration is as originally stated: such assents are necessary means of making something the object of one’s will – that is, of willing an end – rather than merely of coherently willing an end.

It is worth pausing over Kant’s insistence that ‘making something the object of my will’ requires that we must ‘suppose its possibility’ (KpV, 5: 142). Crucially, this implies that one genuinely wills an end – sees it as normatively binding – only if one assents to its attainability. To make something the object of one’s will is to will the world to be a certain way, committing oneself to the idea that one’s actions help achieve it. It is precisely because I see my actions contributing to this attainable state of the world that normative claims have their grip on me. Conversely, without assent to Attainability, normative claims lose this grip. Thus, Kant somewhat overstates his position when he writes that ‘it would be practically impossible to strive for the object of a concept that would be, at bottom, empty and without object’ (KpV, 5: 143; my italics). More accurately, it would be impossible to strive for such an object in a way that generates normative demands. This is presumably Kant’s point when he claims that, were the Highest Good impossible, the moral law would be ‘fantastic, directed to empty imaginary ends, and consequently inherently false’ (KpV, 5: 114). The moral law is ‘false’ here in the sense that it fails to normatively bind us. Let me now unpack this thought.

Central to Kant’s conception of practical reason is the idea that reasoning can guide action because it is structured by maxims. Maxims have specific propositional content: in a particular context, I will adopt a certain means to achieve a certain end. Thus, practical reason, in Kant’s sense, commits us to adopting particular actions as means to particular ends in the world, under specific circumstances.

Thus, to make an end the object of my will is to commit to adopting the means necessary to bring it about.Footnote ²² An example will help to illustrate Kant’s position. If I make fluency in a new language the object of my will and believe that achieving this end requires regular study, then I commit myself to studying regularly, believing that my studying contributes to that end. This commitment also precludes my abandoning my studies without reason. Of course, having taken fluency as the object of my will, I might later abandon my studies to pursue another goal, such as completing a marathon or learning the piano. But if I do so, I have abandoned both the end – fluency in a new language – and the belief that my studying contributes to achieving it. As a result, I no longer have a normative demand to continue studying. I might still believe that studying helps one learn a new language, but I no longer see my studying as contributing to that end, and therefore no longer see the end as providing a normative reason to study.

To be clear, the Kantian framework permits evidence regarding the difficulty – or even impossibility – of attaining a particular end to influence an agent’s practical reasoning, but solely in relation to the initial adoption of that end and the means one should adopt when achieving an end that is difficult. This was partly the point of showing the Kant’s moral argument starts with an independent premise to will the Highest Good. Considerations of impossibility reveal what one should or should not take up as ends. The fact that a spaceship cannot – given the nomological laws – travel faster than light is a reason I should not set building a faster-than-light spaceship as my end. Considerations of difficulty reveal what kinds of means one should adopt to achieve an end. The fact that learning a certain language is difficult is a reason to study when I adopt that end.Footnote ²³ But, once I have adopted an end, I ought to assent that it is attainable because doing so is essential to willing that end.

Read this way, Attainability is a requirement of practical reasoning. Kant holds that practical reasoning is structured by maxims, which function as reasons to act. But for practical reasoning to have normative force, the agent must see their actions (the means) as contributing to their ends. Without assent to Attainability, the agent cannot view their actions as contributing to their ends, because they do not regard a world in which their actions make a difference as possible. In this sense, their maxims lose normative force – they no longer make sense as purposeful steps towards achieving a goal.

4.2 Response to not-believing objections

On the proposed reading, when practical reason guides us, we are responsive to reasons – and this responsiveness has normative force only if we assent to Attainability. If this is correct, then cases like Test Taker are less decisive than Willaschek suggests. For if the relevant rational consideration is not one of structural rationality, then – even though Kant says little about the coherence of our mental states in the Critique of Practical Reason – he can allow that rationally trying to do something may require only withholding judgement about its possible attainment.

The Kantian point, however, is that in such cases one does not make something the object of one’s will, because one fails to take a stance on how one’s actions can contribute to one’s ends. As we have seen, for Kant, in willing an end, I am committed to bringing about that end and to seeing myself as contributing to its realization. Thus, for example, if I do not assent that this world is the kind of world in which passing the test is attainable, then I am not genuinely committed to passing the test. That is, I am not committed to taking up the necessary means – studying, arriving at the venue on time, etc. – required to will to pass it. And so I am not normatively bound to do so.

As a result, Kant’s appeal to a positive attitude in Attainability is justified. Where the not-believing objector goes wrong is in insisting that the relevant rational consideration concerns structural rationality, rather than the conditions that make practical reasoning possible.

At this stage of the argument, one might object that my reading fails to account for Kant’s other practical considerations – specifically, that our assent to God’s existence stems from a ‘need of pure practical reason’ (KpV, 5: 142), and that it is here that rational coherence enters the picture. In response, it is important to distinguish between two practical considerations in Kant’s moral argument: a ‘need of pure practical reason’ and the notion of ‘means’ to practical ends (KpV, 5: 146). On the one hand, Kant holds that we have a need of pure practical reason to represent the attainment of the Highest Good. This need leads to Belief in God, because ‘our reason finds it impossible for it to conceive’ a course of nature in which happiness and morality are connected in exact proportion (KpV, 5: 142). On the other hand, Kant also presents a practical consideration that recommends, as a means to an objectively necessary end (i.e. our willing of the Highest Good), the ‘presupposition of its possibility’ (KpV, 5: 146). Both of these practical considerations are at play in the following passage, partially quoted above:

This passage indicates that there are two distinct practical considerations at work: one grounded in a need, and the other in a means–end relation. This fits well with my view that Kant’s moral argument proceeds in two steps: (1) our willing of the Highest Good as an end provides practical justification for assenting to Attainability; and (2) we are thereby rationally warranted in assenting to God’s existence. Although I remain doubtful about Kant’s arguments for Step 2, my reading accommodates the need of pure practical reason – but only after the practical considerations underlying Attainability have been established.

4.3 Attainability

The second element of the present proposal concerns the content of Attainability. To begin, let us restate Attainability in the form of the following principle:

Whereas Attainability (1) was framed as a constraint on rational willing, Attainability (2) is a constraint on willing simpliciter, reflecting the discussion in Sections 4.1 and 4.2. However, this new principle still falls short of fully capturing the notion of practical possibility. For Kant, a state of affairs is practically possible only if it lies within our power to make it real in this world. In this sense, he claims that something is ‘practical for us’ insofar as it is ‘to be made real through [the freedom of] our will’ (KpV, 5: 113), and he further maintains that we have a moral obligation to will the ‘production of the Highest Good in the world’ (KpV, 5: 122).Footnote ²⁴ Hence:

Attainability (3) contains a modal claim stronger than that a relevant end is attainable merely in a logically possible world. It captures Kant’s claim that Attainability is required to render what is possible – ‘through action’, and therefore, through ‘cause and effect’ (KpV, 5: 113) – ‘actual’ in the ‘world of senses’ (JL, 9: 68). As such, the principle expresses the claim that, once we acquire a rational obligation to pursue an end belonging to this world, we become committed to thinking that we can, indeed, make that end real in this very world. In this sense, practical possibility concerns the background conditions of the world in which our ends can be realised by our free will. Thus, Kant is not claiming that it is rational to think that God will step in to take the necessary actions to realise our ends. Rather, his claim is that in willing our ends, we think that our ends are attainable through our actions. As such, if the Highest Good is practically possible, then we must think that we can produce it, or at least contribute to its production, through our actions, in a world amenable to those actions.

Finally, we must capture the fact that Attainability concerns a subject’s contribution to the complete attainment of a common end, rather than an attempt to bring it about single-handedly.Footnote ²⁵ We must do this because our duty to will the Highest Good is a duty to do our part.Footnote ²⁶ As Kant says, we are to realise the Highest Good through our actions by doing what is immediately ‘within our power’ combined with elements beyond our control that are a ‘supplement to our inability’ (KpV, 5: 119).Footnote ²⁷ Thus:

We are familiar with the notion of the subjects’ individual duty, which describes a duty to attempt to bring about an end on one’s own. Many of our duties belong to this kind, including, for example, our duties not to lie, murder or steal. Presumably, individual agents can realise these ends on their own.

By contrast, a duty to do one’s part is a duty to do something that contributes to attaining some end if it is combined with the actions of others. Duties to do our part provide a compelling account of many of our social and political goals. For instance, it is plausible that we have a duty to end inequality. However, this does not imply that we have a duty to do so on our own, but only as participants of a collective action comprising many individuals.

Kant’s Highest Good likewise involves a division of labour, namely, a duty to do our part in working towards a communal end in combination with others’ efforts. By way of example, consider a soccer player who tries to do their best in every game while knowing that their team will never win the championship. They reason, of course, that this is all that is within their power to do. In this case, however, their end is trying their best irrespective of a broader goal. Admittedly, this does not preclude them from winning the championship by accident, but only that they were willing that end. Similarly, we are responsible for becoming morally worthy of happiness – as that is within our power. God is responsible for the distribution of happiness with respect to morality. A Belief in God is not needed for us to try our best to be morally worthy. Rather, such a Belief is required for us to will to our best ability morally worthy ends, as a component of willing the Highest Good, alongside the distribution of happiness. This picture fits nicely with the way Kant in the Theory and Practice essay, in clarifying his position against his critics, characterizes our duty towards the Highest Good and rational Belief: our duty to promote the Highest Good is a duty to ‘work to the best of one’s ability towards the highest good possible in the world … which, since it is within our control from one quarter but not from both taken together, exacts from reason Belief’ in God (TP, 8: 279; original emphasis).

This division of labour reading has the benefit of assuaging two worries. First, one might worry that Attainability implies that agents are rational in thinking that God will step in to complete our duty. Second, it might be objected that we cannot be responsible for ensuring that morally good agents from the past, who did not receive happiness in proportion to their virtue, are made happy now. This would be beyond the capacity of present agents and thus violate ought-implies-can. However, on the division of labour reading, God is not required to step in and make us worthy of happiness, as that is not God’s role in the promotion of the Highest Good. Furthermore, we are responsible for making ourselves worthy of happiness, not its distribution to past, present or future agents.

4.4 Response to approximation objections

We are now in a position to see where approximation objections go wrong. These objectors argue that, because the Highest Good cannot be fully realized by us, it cannot be true that (A) we ought to will the Highest Good as an end without violating the principle that ought implies can. Therefore, they propose a revised claim: (A_{approximation}) we ought to will an approximation of the Highest Good. However, on the proposed reading, ought-implies-can is not violated. The failure of implication is due to the fact that individual duties differ from duties to do one’s part. The mere fact that we cannot bring about a certain end by ourselves implies nothing about the possibility of that end’s full attainment, except that it will require the combined actions of agents.

Indeed, Kant criticizes merely virtuous (i.e. Stoic) and merely eudaimonistic (i.e. Epicurean) conceptions of the Highest Good because they make the human will ‘the sole and sufficient ground of [the Highest Good’s] possibility’ (KpV, 5: 126). By contrast, the Kantian composite Highest Good is attainable only through our ‘collaboration’ (TP, 8: 47n) with forces beyond our control. Thus, my reading suggests that when Kant says that we have a ‘duty to realise the Highest Good to the utmost of our capacity’ (KpV, 5: 144n), he means that the Highest Good belongs to a class of duties that require us to do our part.

Consider a violinist in an orchestra. She has a duty to do her part in achieving a successful musical performance only if her playing would contribute to the success of the performance along with the combined performances of the orchestra’s other players. Her task is not to play the other musicians’ parts, nor to conduct the orchestra. As such, on her own, she cannot achieve the full performance. But that fact does not imply that she ought only to approximate the successful performance, or that the performance cannot be fully realised. It implies only that the full realisation requires that she engage in a collaborative effort together with her fellow musicians. Her duty is consistent with ought-implies-can: she can dispense her duty to do her part to achieve full realisation. Given that she can contribute to the success of the performance through her actions, she would not be willing her end were she to make this contribution while also thinking that such a contribution to full success is impossible – as Attainability (4) predicts.Footnote ²⁸