1. The aim (and key assumptions) of the argument

It is common enough in modern moral and political philosophy to assume that one can convert an imperfect duty into a perfect one, usually by exploiting aspects of specific cases to determine the classification.Footnote ¹ Now, whether one can or not will depend on which criteria of perfect and imperfect duty one is using. I shall say nothing here about perfect/imperfect duty distinctions that either pre-date or post-date Kant.Footnote ² With respect to Kant’s distinction, a moment’s reflection should reveal the impossibility of conversion. Kant’s core criteria of perfect/imperfect duty are given at G, 4: 424: perfect duties are contrary to that which is contradictory in conception; imperfect duties contrary to that which is consistent in conception but nevertheless contradictory in will.Footnote ³ This is a formal distinction in kind, so the idea of conversion must be absurd: that which is contrary to what is consistent in conception cannot be contrary to what is contradictory in conception.

And yet, when one digs deeper into Kant’s account of perfect and imperfect duty, some unexpected results occur. At CPrR, 5: 159, Kant distinguishes duties that are essential/non-essential when it comes to living up to the moral law, and at MM, 6: 390, he distinguishes duties that do/do not admit of degrees of virtue or goodness. Kant’s examples and terminology, in both passages, clearly point back to the perfect/imperfect distinction: these are supposed to be aspects of the core distinction. However, when we attend to Kant’s late distinction between duties and their token obligations, we uncover cases where some imperfect obligations operate in a perfect-ish way, and some perfect obligations operate in an imperfect-ish way. As we will see, while all imperfect duties are non-essential, some imperfect obligations may be essential. It is also possible, I shall argue, for some perfect obligations to admit of degrees of virtue: because Kant casts his distinction negatively, in terms of rejecting principles that generate contradictions, the possibility obtains of someone carrying their rejection of that which is contrary to perfect duty further than they might have – in a way that admits of degrees of virtue or goodness. None of this amounts to the possibility of conversion, exactly. But the picture is now sufficiently complicated that its coherence needs some explaining.

The primary aim of this article is to offer an explanation that preserves the coherence of Kant’s account: there is no inconsistency, only a remarkable degree of complexity and subtlety. To show this, I first offer (sections 2-4) an interpretation of the central practical difference of the G, 4: 424, criteria: imperfect duties exhibit extra demands on judgement (concerning who to enact duties for, and how to bring the demands of duty into a unity in one’s activity) that perfect duties lack. I then use this account of latitude to explain how both the possibility of an essential imperfect obligation, and of a perfect obligation admitting of degrees of virtue, are consistent with Kant’s core criteria. In the former case (section 5), the possibility is explained by how latitude may play out depending on context. In the latter case (sections 6–7), the possibility is explained by how perfect obligations can only admit of degrees of virtue above a certain threshold of conduct: perfect and imperfect duties thus exhibit subtly different connections to virtue or goodness. I conclude (section 8) by showing how this complexity enables a promising Kantian response to concerns over the classification of the duty of beneficence.

In making this argument, I will be relying on four main interpretative assumptions. First, I assume that a practical account of the contradiction in conception/will tests, broadly along the lines Christine Korsgaard and Onora O’Neill have proposed, is correct.Footnote ⁴ Second, I assume Kant does not abandon the Groundwork criteria of perfect/imperfect duty in later work: I read the later passages not as revisions to, but as fleshing out aspects of, the core Groundwork distinction.Footnote ⁵

Third, I take Marcus Willaschek to have shown that Kant’s perfect/imperfect distinction is not, despite the impression Kant himself sometimes gives, co-extensional with Kant’s right/ethics distinction: all duties of right are perfect duties, but not all perfect duties are duties of right (Willaschek Reference Willaschek1997: 207–8). So we can consider the perfect/imperfect distinction in some degree of isolation from the right/ethics distinction, though the latter should certainly be on our radar when considering the former.

Finally, I here assume that the latitude imperfect duties exhibit contains no permissive element. One cannot say, with reference to nothing other than the duty itself, ‘I’ve done enough for now, I need do no more’ (i.e. without needing to reference other moral demands that take precedence). This assumption flatly contradicts Thomas Hill Jr’s sustained argument for a permissive element within Kant’s criteria (e.g. Hill Reference Hill2002). I am not going to defend my assumption here,Footnote ⁶ because I hope I can set this controversy aside. What matters, for my purposes, is that we understand imperfect duties’ latitude as placing extra demands on judgement. A defender of Hill’s permissive interpretation could agree with that, and add ‘But latitude grants permissions for judgement as well’. It therefore strikes me that a defender of the permissive interpretation need not find anything to disagree with in the key move that I will make, and the controversy can be set aside.

On the above points, then, I shall be arguing from rather than for a particular interpretation of Kant. So much for what I will not try to say – it is now time to get on with the argument.

2. The perfect/imperfect distinction

In the Groundwork, Kant distinguishes perfect from imperfect duty as follows:

Against some interpretations, I take the passage to be definitional.Footnote ⁷ One of Kant’s purposes in this passage is to take terminology favoured by his philosophical opponents – in this instance, early modern natural law theorists – and recast it within his own philosophical system while preserving something of its popular meaning. Early modern natural law theorists distinguish imperfect duties to enact God’s love on earth through Christian charity from perfect duties of human justice.Footnote ⁸ That distinction forms part of the natural lawyers’ project of naturalizing the Thomistic categories of divine and natural law, reframing morality in terms of human sociability rather than our complex movement towards God. Kant is redefining the perfect/imperfect distinction, supplanting the natural lawyers’ criteria with his own.

Two observations – amounting, alas, to circumstantial evidence at best – lead me to this interpretative suggestion. One is that it is clear that Kant’s engagement with the natural law tradition is sustained and complex.Footnote ⁹ And second, Kant certainly avails himself of this strategy elsewhere. Jens Timmermann has shown how the three variant formulae of the categorical imperative are introduced to recast various tenets of natural law and Leibnizian ethics (Timmermann Reference Timmermann2007: 78–106). We might also note how, in the first Critique, the Platonic term ‘Ideas of reason’ is recast against Platonism (O’Neill Reference O’Neill2015: 18). Of course, evidence of a general tactic on Kant’s part is not evidence that he is deploying that tactic in the Groundwork definition of perfect/imperfect duty. Yet I find it very hard to credit the idea that, despite his sustained engagement with the natural law tradition on so many points, Kant did not have its division of duties centrally in view as he offered his own.

If I am correct that the passage defines the criteria of perfect/imperfect duty, two further points are crucial. First, Kant’s criteria are entirely negative. Perfect duties are contrary to principles that are contradictory in conception. Imperfect duties are contrary to principles that are consistent in conception, but contradictory in will. Following Onora O’Neill’s suggestion, I take this to mean that principles of duty are principles that reject principles containing the relevant contradiction (O’Neill 1996: 165–6; Reference O’Neill2015: 48).

This immediately leads to a second point. Kant’s distinction between perfect/imperfect duty is a distinction in kind: that which is contradictory in conception cannot be that which is consistent in conception but nevertheless contradictory in will. But it is not a distinction of opposites (as e.g. between ‘positive’ and ‘negative’ duties). Imperfect duties, unlike perfect duties, reject principles that pass the contradiction in conception test. But both sides of the distinction reject principles that fail the contradiction in will test, because generating a contradiction in conception is one (but not the only) way a principle generates a contradiction in will.

Digging deeper, these tests concern the same sort of contradiction: a principle that claims inclusivity for itself is in fact exclusive. This does not collapse the two tests into one, for they pick out two different locations for this sort of contradiction. In Stephen Engstrom’s acute formulation, a contradiction in conception is located in what is willed; a contradiction in will is located in the willing itself (Engstrom Reference Engstrom2009: 230). We could put this another way: a contradiction in conception lies in the practice, in what it would be for an unbounded manifold of agents to live by that principle; a contradiction in will lies in the reasoning by which such agents are to adopt the principle as a principle to live by.

A principle is contradictory in conception when it both subverts and upholds a general commitment: with respect to duties to others, when I exempt myself from a commitment that I intend others to honour. A maxim of false promising, for example, both subverts (I am not bound by my promise) and upholds (you must take my promise as genuine) the commitment to keeping promises. This is also the case with, for example, the thief who steals something to own it, or the torturer who exercises their agency to disrupt or destroy the agency of others.

A point of scope is central here. Note how the principle of theft as a mode of acquisition generates no contradiction if the victim is thought of merely as a body on which the participants act out the practice that manifests this principle. The contradiction arises when we ask whether the principle is lawful – that is, a principle for all. When that is our question, the victim must be seen as a fellow participant in the practice of acquisition – something which cannot be consistent with the way in which the principle of theft necessarily excludes the victim.

Turning now to imperfect duty, the contradiction is located not in what is willed, but in the willing itself. To see this, it may help to take a step back. Because any finitely rational agent is capable of adopting as maxims any principles they see fit, to will a maxim I must, minimally, see some reason to do so. Accordingly, the idea of willing a maxim without contradiction must pick out something about the reasoning by which I adopt the principle as a maxim. (After all, ‘the will is nothing other than practical reason’, G, 4: 412.)

The same point of scope is again central: we are thinking (with respect to duties to others) of what must be true of a principle if it is to be a principle for all. If the reasoning by which I adopt a principle as a lawful maxim is predicated on assumptions about agents (their dispositions, outlooks, abilities, etc.) that do not hold for rational agents without exception, then my reasoning will have nothing to say to those for whom my assumptions do not hold, and they can have no grounds for judging, as I do, that this is a principle to live by. My reasoning will be heteronomous rather than autonomous (O’Neill Reference O’Neill2015: 47). For example, even though I am a needy subject myself, I might rationalize a maxim of never assisting others through some sort of special pleading: ‘my needs matter, the needs of others don’t’. But this must foreclose the possibility of others adopting the principle as a maxim by the reasoning that I adopt it (G, 4: 423; MM, 6: 453). A contradiction now emerges: my maxim of never assisting others is advanced as lawful, yet can only be adopted on grounds that foreclose the possibility of it being lawful. So, just like principles of perfect duty, principles of imperfect duty are inclusive. But the inclusive aspect of the latter lies not in the practice manifesting the principle, but in the reasoning by which the principle is adopted: that reasoning is autonomous rather than heteronomous.

The exclusionary nature of a principle that is contradictory in conception entails that it also involves a contradiction in will: the reasoning by which I could adopt a principle that is contradictory in conception must also involve some kind of special pleading (your stuff can be mine, but my stuff can’t be yours, etc.) that does not meet the publicity conditions of the contradiction in will test. But it is not the case that every principle which generates a contradiction in will generates a contradiction in conception: a shared way of going on in which no one assisted anyone else, for example, does not at once subvert and uphold a general commitment (G, 4: 423).

3. The key practical difference between perfect and imperfect duties to others

Because Kant’s criteria of perfect and imperfect duty test for contradictions in different locations, the key practical difference between perfect and imperfect duty is this: imperfect duties exhibit extra demands on judgement that perfect duties lack.

Perfect duties require that we reject principles that are contradictory in conception. If I am to reject the principle of theft as a mode of acquisition, I must commit to a principle that is not exclusionary in the same way, a principle that is genuinely inclusive. As the contradiction is located in what is willed – what I have called the practice – the relevant perfect duty must be manifest in a practice that is genuinely inclusive, that an unbounded domain of agents can live by without excluding anyone. Two conclusions follow. First, principles of perfect duty to others are principles that each agent can conjunctively enact for all: the question of who to enact a perfect duty for is foreclosed.Footnote ¹⁰ And second, because the relevant practice is genuinely inclusive, the demands of any perfect duty can be construed negatively, making it possible to unify the demands of perfect duties in one’s activity without complication.

By contrast, nothing in the idea of a principle rejecting that which is consistent in conception but nevertheless contradictory in will entails that these questions are foreclosed, because the inclusive element of Kant’s criterion of imperfect duty instead concerns the reasoning by which the duty is adopted. Nor can these questions be otherwise foreclosed, at least for agents with finite powers of thought and action. All principles of duty are prescriptive and prospective, orienting one’s ongoing activity as it plays out across indeterminate possibilities for interaction with an indeterminate number of others. The contradiction in will test tells me, for example, that I must assist others in need, but there can be no guarantee – from the criterion of duty here – that I can conjunctively do so for all I might assist. Nor can there be any guarantee that I can fit every possible manifestation of the set of imperfect duties into the shape of my life. Accordingly, the idea of imperfect duty entails extra demands on practical judgement: I must decide who to enact an imperfect duty for (when a duty to others), and how to fit this duty into a unity in the shape of my life with the other duties I am under.Footnote ¹¹

The extra demands on judgement that imperfect duties exhibit but perfect duties lack also entail that imperfect duties to others cannot be correlative to rights. I here assume that the holder of a right is wronged if the correlative duty is not performed.Footnote ¹² The extra demands on judgement that imperfect duties entail foreclose correlativity. Because I must select for whom I enact an imperfect duty, it cannot be the case that any one possible recipient has a claim that they, rather than someone else, be assisted. Here, for example, are Sean, Vaughn and Bjorn, all stranded on separate rocks against a rising tide. I have only enough time to rescue two of the three before the tide claims them (assume none stand in any special relationship to me). In rescuing as many as I can, I am doing exactly what morality requires of me. I do not wrong Bjorn if I rescue Sean and Vaughn, so Bjorn cannot have had a right that he be assisted. And if Bjorn has no claim that I assist him rather than the others, neither do Sean or Vaughn.

4. Two loose ends: duties to self and duties of right/ethics

At this point, I had better pause to tidy up two loose ends. You will have no doubt noticed that the discussion so far has concerned only duties to others. To be very brief about duties to oneself: the same structure applies, but temporal considerations replace considerations of a manifold of agents.Footnote ¹³ That is, the relevant idea of inclusivity no longer concerns a plurality of others (how could it?) but the judging agent’s own activity extending into an open-ended future.

Thinking as a duty-bearer, a duty is a certain kind of practical end: something one acts for the sake of. Note, further, that this kind of end has a certain generality to it. It is time-general. When I, say, buy some butter or fetch my camera, my end is something I bring to completion: if in future I decide to buy butter again, I am setting myself a new end. Maxims of duty, by contrast, are not ends that come to completion in discrete patterns of action. When I lift a floundering baby from a pond, and then check in on my elderly neighbour upon arriving home, I am in each case acting for the sake of beneficence. But I am not setting myself a new end each time. I am manifesting the same end on multiple occasions.Footnote ¹⁴ Because a duty, so understood, orients my ongoing activity and has this time-general form, I must understand the duty as binding at all times. With respect to duties to oneself, the exclusionary aspect concerns my own action in time: perfect duties are contrary to principles that at once affirm and subvert their manifestation in my ongoing activity; imperfect duties are contrary to principles adopted by reasoning incompatible with the adoption as maxims of the other principles I must live by.

The second loose end concerns how Kant’s criteria of perfect and imperfect duty relate to the right/ethics distinction. Duties of right, in contrast to duties of ethics, are appropriate objects of legitimate coercion, and the possibility of legitimate coercion presupposes the existence of an omnilateral will that is external to any subject’s will, individually or collectively (e.g. MM, 6: 263). Duties of right thus concern only the external action of the duty-bearer, that my action conforms with the requirements of right (6: 231). This is because, as an appropriate object of legitimate coercion, a duty of right is something the sovereign enacts through her citizens.Footnote ¹⁵

For our purposes, two quick observations follow regarding how the idea of a duty of right, and of a perfect duty, match up. One, we can see how duties of right are perfect duties, because Kant’s criterion of perfect duty locates the relevant contradiction in what is willed, in external action. But – and this is the second observation – there is no obvious reason to assume that all duties contrary to contradictions in conception are appropriate objects of legitimate coercion. So we should not assume all perfect duties exhibit all the features of duties of right: duties of right are a subset of perfect duties, perhaps even a subset of perfect duties to others.Footnote ¹⁶

5. Essential and non-essential duties

I turn now to our second passage, from the Critique of Practical Reason:

The examples here clearly point us back to the Groundwork criteria of perfect/imperfect duty. And once again, natural law terminology is being recast. As part of the project of naturalizing Thomistic natural/divine law, the early-modern natural lawyers and their descendants made perfect duties cover that which is essential for human sociability, and imperfect duties cover that which merely improves or aids human sociability.Footnote ¹⁷ If I understand him, Kant is recasting ‘essential’ as the necessity of enacting the duty in question if one is to live in accordance with the moral law (henceforth: living well), not in terms of human sociability.

With respect to some duties, one cannot be said to be living well if one passes up an opportunity to enact them. With respect to other duties, one can still be said to be living well even if one has passed up an opportunity to enact the duty in question. That possibility is a function of the extra demands on judgement exhibited by imperfect duties that perfect duties lack. Because I must judge who I enact an imperfect duty for, and I must also judge how to best integrate all the imperfect duties with the demands of perfect duty in the shape of my life, there will always be cases where I pass up an opportunity to manifest a particular imperfect duty in favour of either a different manifestation of the same duty or for the sake of some other duty. In these cases, I am still living well. I am still doing exactly what morality demands of me.Footnote ¹⁸ So perfect duties are essential, and imperfect duties non-essential.

Yet matters are not so clean cut! To see this, we need to think in terms of Kant’s late technical distinction between duties, the lawful principles one should live by, and obligations, the particular tokens of duties in this or that situation (MM, 6: 222). The essential/non-essential distinction matches the perfect/imperfect distinction at the level of duty (Pflicht). But the same is not true at the level of obligation (Verbindlichkeit).

Any token of a principle contrary to that which is contradictory in conception cannot be contrary to that which is consistent in conception: perfect duties generate perfect obligations while imperfect duties generate imperfect obligations. But it may be the case that some obligations are essential even though they are tokens of imperfect (and thus non-essential) duties. I have in mind something Marcia Baron once said: there are cases where failing to enact the duty is such an egregious moral error that we could not seriously say, of someone who makes the error, that they are committed to acting for the sake of the duty at all.Footnote ¹⁹ If, for example, I could easily save your life by calling 999 as you collapse in front of me, but I choose not to, there is surely no genuine sense in which I understand a principle of assisting those in need as a maxim to live by. The duty here is non-essential (there will be instances where my passing up the opportunity to assist others does not mean I am failing to live well) but the obligation is essential (this is not one of those instances).

There is an asymmetry here: while a non-essential duty can generate an essential obligation, an essential duty cannot generate a non-essential obligation, precisely because an essential duty does not exhibit the relevant demands on judgement. We can certainly imagine cases where one might forgo a perfect obligation for the sake of another perfect obligation: for example, where all my options involve violating others’ rights. But such tragic cases are not of the correct sort. They are cases where whatever I do I do wrong, so my passing over the obligation does impinge on my living well.

If all this is correct, then the essential/non-essential distinction is not straightforwardly co-extensional with the perfect/imperfect distinction. Yet the former is still an aspect of the latter. Both distinctions are distinctions in kind, but only the perfect/imperfect distinction is a formal distinction. Whether or not an obligation is essential for living well is instead contingent on the circumstances the duty-bearer is in.

6. Duty and virtue

I turn now to our third passage. In the Metaphysics of Morals, Kant tells us:

Timmermann has recently drawn a helpful link between this passage and an earlier draft of the Doctrine of Virtue. As the draft tells us:

This link constitutes good grounds for reading Kant’s claim about the latitude imperfect duties exhibit in terms of extra demands on (rather than permissions for) judgement that perfect duties lack (see also MM, 6: 411). Judgements about how to manifest the demands of an imperfect duty, and to what extent one should do so, are judgements that concern how to manifest the moral law in the situation one is in, rather than whether one feels one has done enough in general. The claim about latitude thus makes explicit the aspect of the Groundwork criteria I was concerned to bring out in section 3.

Yet there is more going on in the passage. One of Kant’s concerns, in the surrounding text, is to distinguish duties of right from considerations of virtue or goodness. The passage connects that point to the perfect/imperfect distinction: imperfect duties alone concern ‘the field for the practice of virtue’, and so can admit of degrees (‘the field … of virtue is widened’). So we have a distinction between duties that admit of degrees of goodness or virtue and duties that do not, that is supposed to map onto the perfect/imperfect distinction.

My impression is that this distinction is obscured by a relatively common gloss on the passage, which misplaces the relevant scalar dimension by distinguishing duties to perform actions from duties to adopt ends, as if the distinction concerns the degree of precision with which duties are specified. I tentatively suggest that this gloss is misled by how the point in the passage concerning the perfect/imperfect distinction is somewhat blurred with the points concerning the right/ethics distinction Kant is in the process of making (indeed, Kant here straightforwardly equates narrow duties with duties of right). Kant is certainly clear that duties of right, unlike ethical duties, have ‘a precision analogous to that of mathematics’ (MM, 6: 375n.). If I understand him, that is because duties of right are appropriate objects of legitimate coercion: all the sovereign can legitimately coerce us to do is to follow precisely the letter of the commands she issues. But it would be inappropriate to simply transpose this difference between duties of right and ethics onto the perfect/imperfect distinction.

One reason it would be inappropriate is that there is a clear sense in which principles of both perfect and imperfect duty are ends: maxims of duty feature in practical thought as time-general ends, standards that orient one’s ongoing and open-ended activity as that plays out across an indeterminate number of possibilities for interaction with an indeterminate domain of agents. In this regard any perfect duty, just as much as any imperfect duty, is a duty to adopt an end that does not come to completion in any specific pattern of action.

Furthermore, there is also a clear sense in which both perfect and imperfect duties precisely specify what is to be done. Recall that the demands of perfect duty can be construed negatively: thus, while there are an indefinite number of ways to murder someone, the only way to manifest the perfect duty rejecting murder is to never do any of them. That is about as direct as one can get! Yet the imperfect duty of beneficence is also direct. Human needs are kaleidoscopically complex, and they may be met in an indefinite number of ways. But if I am to do anything about any such needs I must, in any situation I am in, to the best of my ability, identify the needs of others that call for a response. That is a direct specification of a duty that requires us to help ‘where one can’ (G, 4: 398).

All we are left with is a mere difference of degree. To focus again on the duty of beneficence, even though it provides a direct specification, it is often going to be the case for token obligations that the duty-bearer must make a considerable number of judgements shaping the concrete pattern of action that successfully manifests beneficence. Think, for example, of what to do when one discovers that one’s elderly neighbour, disconnected from any family, is developing Alzheimer’s disease. But there is no guarantee that this complexity will be true of every imperfect obligation. What to do when you see someone caught in a riptide, and the lifeguard team is in easy reach, is hardly complex! And we can identify perfect obligations that are more complex than this second example. For most of us, the perfect duty prohibiting murder is straightforward. But what to do will be much less straightforward for, say, a newly patched Mongrel Mob member charged with placing a hit on a rival gang and constantly under the distrustful, watchful eyes of his gang brothers.

If we instead read the 6: 390 passage as pointing to how considerations of virtue or goodness fall only on the imperfect side of Kant’s perfect/imperfect distinction, we do not saddle Kant with trying to make a difference of degree stick to a distinction in kind it cannot map onto. Indeed – and again, following Timmermann’s recent lead – the connection to virtue appears to provide a subtle way of reconciling the scalar and non-scalar aspects Kant has in view. Precisely because imperfect duties entail extra demands on judgement that perfect duties do not, it is possible – on the imperfect side only – to speak of acting from duty in scalar terms. I can act well in manifesting the demands of imperfect duty in what I do, while acknowledging the possibility that I might have acted, or in future act, better in terms of how I consider and respond to those demands (Timmermann Reference Timmermann2018b: 389ff.). Citing Kant’s note that ‘a good action could be better, but a right one not be righter’ (R 7036),Footnote ²¹ Timmermann takes it to be clear that considerations of virtue or goodness apply only to imperfect duty.

7. Perfect obligations may admit of degrees of virtue

Yet once again, matters are not so clean cut. The difficulty here is that Kant’s negative criteria of perfect and imperfect duty leave room for considerations of goodness falling on both sides of the perfect/imperfect distinction. One right act cannot be righter than another, but it might be better.

To see this, consider a general observation O’Neill makes concerning supererogation. Against the common understanding of supererogatory and obligatory acts as involving different principles, O’Neill observes that ‘[s]aints and heroes are admired because they fulfil quite ordinary ethical requirements, but do so with superabundance … The saintly or heroic act was indeed a token of a required type of action, but a token that exceeded all expectations and all ordinary measures of duty’ (O’Neill Reference O’Neill1996: 207–8). The obligatory act and the supererogatory act, on O’Neill’s picture, are instances of acting on the same principle.

As what is true of superabundance is also true of abundance, I shall set aside supererogation. By Kant’s negative criteria, principles of perfect duty are principles that reject principles that are contradictory in conception. The notion of rejection, here, makes it possible for considerations of abundance to apply on both sides of the perfect/imperfect distinction. When I commit no murders and tell no lies, I am acting for the sake of perfect duties rejecting violence and deception. But when I also tirelessly work to expose and correct the systemic underpinnings of racially motivated uses of lethal force by my city’s police department, or to expose and correct the false promises and misinformation made during an election or referendum, I am also acting for the sake of perfect duties rejecting violence and deception. These latter courses of action may be incredibly time-consuming, difficult and in many contexts risky. They go beyond the standard measure of the relevant perfect duties and seem clear cases of enlarged virtue.Footnote ²² Kant’s own discussion of the parental duty to care for one’s child so far as one can (MM, 6: 281) is another instance of a perfect duty that plausibly has room for meritorious conduct, depending on how far one carries it forward.Footnote ²³

You might well ask: why should we think of these examples as token obligations of the same duties? If there are different duties here, we need a systematic way of distinguishing them. It is not clear to me one can be found. The most intuitively promising seems to be as follows. When I commit no murders and tell no lies, I am ensuring that my own conduct does not fall short of what morality requires. But when I take on the institutionally racist police department or the dishonest politicians, I am responding to others’ wrongdoing. The duty to respond is distinct, in both principle (protecting the innocent perhaps, rather than a rejection of violence) and in content (it may, in some cases, limit other duties).

The difficulty with this imagined reply is that the question ‘How should I live?’ already entails a manifold of others: not in the pronoun, but in the idea of living, which can only be the idea of interactions with others who are themselves interacting with me. Because what I do plays out in interactions both with others and with the institutional structures through which those patterns of interaction unfold, any sharp distinction between my conduct and my reactions to others’ conduct should surely strike us as forced: what I do is always in some sense what we do. This should be heard as denoting a sphere of institutionally mediated interactions, bounded in different ways with respect to different institutional contexts, and existing in a complex interrelation with other spheres that I might possibly enter: the relevant ‘we’ is not an unbounded one. That is enough, I hope, to put the possibility of complicity in others’ actions or purely institutional injustice squarely into the very idea of my own activity.

I do not mean to suggest that the idea of rejecting violence or deception should be understood as bringing the total amount of violence or deception down to zero, as a consequentialist might say. The suggestion, rather, is that in acting for the sake of perfect duties rejecting violence and deception I cannot be inattentive to the institutional structures my life is lived through. After all, the personal is political – a point Kant himself seems somewhat alive to in his example of the man who, recognizing he is the beneficiary of injustice, is motivated to perform great acts to redress this (CPrR, 5: 155n.).

Here we find a distinctive element of latitude involved in what it would be to reject violence or deception. As noted earlier, one can construe the requirements of perfect duty entirely negatively: I commit no murders and tell no lies, and also refrain from various forms of direct complicity in the violence or duplicity of other natural and artificial agents. Yet if I am also attentive to the ways in which wrongs are frequently complex, institutionally embodied and interconnected, I must understand my responses as contributions to patterns of action that are in various ways collective. That opens up scope for carrying my rejection of violence and deception further, in the extent to which I engage with the collective dimensions: for example, engaging with the practical need, when it arises, for more organised or structured collective action. To put it crudely – but not, I hope, trivially – rejecting violence means one must join the fight against one’s institutionally racist police department, but one does not have to be the movement’s treasurer or the person who delivers the submission to Parliament. This is something that admits of degrees, and it is far from clear that I do wrong if I do not carry this engagement as far as I can. On the other hand, carrying it far would be a manifestation of great virtue. These token obligations are not obviously tokens of imperfect duties rejecting principles that are consistent in conception but contradictory in will. In carrying my engagement forward I am still acting for the sake of the same principles of perfect duty I act for the sake of when I commit no murders and tell no lies.

At this point, we seem to have reached a conclusion that should not be possible by Kant’s criteria of perfect/imperfect duty. Because Kant’s negative criteria allow considerations of virtue or goodness on both sides of the distinction, it follows – does it not? – that perfect duties may exhibit at least some of the latitude imperfect duties exhibit. That should not be possible by Kant’s criteria!

The best resolution, I suggest, is to bring out exactly what is distinctive about the latitude perfect obligations may exhibit. The most effective way I can think to do this is by first considering how attentiveness to the myriad ways in which the personal is political plays out on the imperfect side of the distinction.

The duty of beneficence, Kant tells us, is the duty to help those in need ‘where one can’. That, too, must require attentiveness to the institutional structures one lives under, for no institutional scheme is seamless. Needy individuals may fall through the inevitable cracks in an institutional scheme, or live in its blind spots, for all sorts of complex reasons. Think, for example, of the plight of the isolated elderly in our societies and the kaleidoscopically complex factors that may cause that isolation: that, for example, cause the fabric of family, friendship and other interpersonal networks to be eroded or destroyed. Helping where one can, here, must mean either entering into or creating, and then sustaining, interpersonal networks, because the very nature of the plight means it will often only be visible through such networks. There is clear scope here for carrying one’s engagement further than one might have, and thus for greater virtue or goodness. But the guiding duty here involves the rejection of a principle which is consistent in conception – the principle of no assistance generates no contradiction in conception – and so this sort of engagement is clearly not something one can possibly do for an indefinite number of possible recipients. The duty thus involves demands on judgement regarding who to assist, and how to unify these requirements with the other principles of duty that orient how one lives.

With this example of an imperfect duty before us, note the signal difference with our earlier cases of apparently meritorious perfect duty. In our cases of meritorious perfect duty, there is no possibility, consistent with doing what morality requires of me, of passing over the duty in question in favour of a different duty, or even in favour of manifesting the same duty in a different context (I do nothing about the police department, while rejecting violence elsewhere). I must, given the institutional racism of my city’s police force, respond to that particular injustice/immorality. So latitude, here, applies only with respect to certain aspects of an essential, perfect obligation. In this way, the latitude a perfect duty can exhibit is strikingly distinct from the latitude imperfect duties exhibit. In the former case, that latitude does not extend to what we might think of as the threshold of duty: to the question of whether the duty can be passed over and one is still said to be living well.Footnote ²⁴ The latitude perfect duties may exhibit only concerns behaviour above the threshold – that is, it concerns some aspects of what it would be to discharge a token obligation of a perfect duty in the situation before one: of what it would be to manifest a perfect duty in a particular pattern of action.

It follows that, although considerations of virtue or goodness cannot be restricted to the imperfect side of the perfect/imperfect distinction, this implication of Kant’s negative criteria of perfect and imperfect duty does not create any inconsistency. Kant’s perfect/imperfect distinction is not unstable but is, rather, remarkably fine-grained.

8. An application: controversy over the duty of beneficence

So, where does this leave us? With, I think, intricate answers to several important questions. How should we understand the differing demands on judgement different duties may place on practical reason? Kant’s negative criteria of perfect and imperfect duty yield a sophisticated answer firmly located within his philosophical system.

How should we understand the possibility that some moral requirements may be passed up even when one is living well? Kant offers us another sophisticated answer, combining sensitivity to context with his systematic classification of categories of duties.

Finally, how should we understand the scalar elements of moral requirements, of virtue or goodness? Kant’s perfect/imperfect distinction contains the resources to give a sophisticated answer, one that takes into account how considerations of merit operate when one thinks, in a way that includes all the complex ways in which living is institutionally mediated and thus political, of what it is to live by a maxim of duty.

In closing, I want to show how this disentangling exercise might be worthwhile, by applying it to the current controversy over the duty of beneficence. In modern moral and political philosophy one finds a remarkable moral certainty that the most important claims of need or assistance are correlative to rights. This claim, often asserted without argument, is, for example, prevalent in the global justice literature.Footnote ²⁵ It is also detectable in some modern Kantian writing on the duty of beneficence. The concern here seems to be that, unless we find a Kantian way past Kant’s own insistence that beneficence is a principle of imperfect duty, Kantians cannot accord certain claims of need the moral seriousness they deserve.

Thus Barbara Herman distinguishes different kinds of duty of assistance. Duties of mutual aid with respect to the necessary conditions for rational agency are separated from duties of mere kindness or help or charity. The former, but not the latter, can be claimed by the subject in need, so (though Herman does not quite put it like this) are perfect rather than imperfect duties (Herman Reference Herman1993: 61).

Alternatively, Karen Stohr argues that beneficence is really a composite of two duties: an imperfect duty to assist and a perfect duty not to be indifferent to others as ends-setters. There will be occasions – often where the needs in question are of the highest moral significance – where my passing over an opportunity to assist is also an expression of indifference to the plight of the needy. As Stohr puts it, ‘although we are not always required to help, we are always required not to be indifferent. When the only way not to be indifferent is to help, we are required to help’ (Stohr Reference Stohr2011: 62).

Both of these solutions strike me as problematic. Stohr’s solution is undermined by the observation that a principle rejecting indifference is contrary to that which is consistent in conception yet contradictory in will: when Kant insists that indifference to the needs of others is consistent in conception, he does not make any exception for indifference to a particular subset of others’ needs. Nor is it clear to me how a Kantian could successfully ground such an exception. As O’Neill notes, ‘[i]ndifference and neglect are not unjust. Nobody can avoid being indifferent to and neglecting many, if not most, others to whom they are connected’ (O’Neill Reference O’Neill1996: 194). The problem here is that Stohr’s duty not to be indifferent to others cannot be the right shape for a perfect duty: it is not a duty I can conjunctively enact for all, so the relevant contradiction cannot lie in what is willed.

Herman’s solution is problematic in a different way. Because there is a gap between need and aid – I need food, but I may require no assistance to acquire it – a duty of aid does not follow straightforwardly from the identification of any need (Engstrom Reference Engstrom1986). Now, perhaps that gap could be bridged. The question is whether there is a recognizably Kantian way of bridging the gap that delivers a subset of duties of assistance that are, as Herman insists, claimable.

Notice how the gap between need and aid has a special resonance within Kant’s position. The gap entails that no one’s agency is necessarily undermined by a principle of never assisting anyone in need: such a principle is not contradictory in conception. Herman rightly argues that no one could expect to get through life without assistance from others, but that thought only shows why a principle of never assisting anyone, although consistent in conception, cannot be universally willed. It does not yield Herman’s claimable duty of mutual aid, for the following reason. A duty that is claimable is such that the recipient is wronged by dereliction of the duty. And, as we have seen, the extra demands on judgement that imperfect duties to others exhibit foreclose the possibility of a potential recipient being wronged by the non-performance of an imperfect duty (recall Sean, Vaughn and Bjorn, stranded on their respective rocks). For the duty to be claimable, it would have to be contrary to that which is consistent in conception.

In modern moral philosophy there is at least one venerable way of bridging the gap between need and aid: the family of interest theories of rights hold that morally salient interests place directed duties on others. But that bridge depends on a specific structure: duties, on almost all interest theories, are derived from a prior conception of well-being. My hunch, and it is only a hunch, is that Herman is covertly operating with exactly that structure, substituting a conception of agency in place of well-being. That seems alien to Kant’s philosophy. The problem here is not just the welfarism of interest theories. Kant’s recursive philosophical method does not derive duties from anything other than what would have to be true of moral reasoning if it is to guide us, so substituting a prior conception of agency still does violence to Kant’s system.

It strikes me that the most effective and straightforward Kantian response to the charge that classifying the duty of beneficence as imperfect downplays the stringency and importance of duties of aid is to emphasize just how fine-grained the perfect/imperfect distinction is. For any constellation of injustices as complex and interconnected as those the literature on rights to aid are concerned with, an adequate response must involve a constellation of moral considerations. Because thinking of the ways in which both perfect and imperfect duty connect to virtue or goodness is, among other things, to think of the myriad ways in which the personal is political, Kant’s division of duties allows us to make sense of a cluster of both perfect and imperfect duties as the appropriate response.Footnote ²⁶

Furthermore, because the distinction between essential and non-essential moral requirements comes apart from the perfect/imperfect distinction we can say, with Baron, that there will be cases where failure to respond to the plight of the needy is so gratuitously wrong that it cannot be said of anyone who does so that they have adopted the principle of beneficence. We can do so because Kant’s negative criterion of imperfect duty does not entail that all imperfect obligations are non-essential. A Kantian can condemn such gratuitous failures in the strongest moral terms while holding that the duty to be beneficent is imperfect.

To this you might object: ‘But if both duty and obligation are imperfect, we still cannot say that the failure wrongs those whose needs are not met.’ That is indeed true. But it is unclear to me why we should accept the assumption, so prevalent in modern moral and political philosophy, that the most important moral considerations are exhausted by a set of basic rights. And it is even less clear why a Kantian should accept it. After all, imperfect duties are contrary to maxims that are contradictory in will, and what could be more central or important to Kant’s practical philosophy than those maxims which can be consistently willed by all?