1. Kant, Constructivism and Contextualism

Rawls’s reading of Kant as a constructivist is among his most enduring legacies.Footnote ¹ It is not easy to come up with a general specification of constructivism; there are many different variants, and it is not clear what, if anything, unites them. For some, constructivism is a development out of contractualism; relatedly, it is sometimes cast as a distinctly democratic decision procedure that secures formal agreement on substantive principles much in the way democratic voting does. Others treat constructivism as a more general theory or method of normative justification, contrasting its bootstrapping generation of valid moral principles with the objectivism of moral realism, though there is disagreement as to whether constructivism is in fact best characterized as a meta-ethical theory at all or whether it is better seen as an alternative to meta-ethical justification. Yet a third approach casts constructivism as a distinctly first-personal perspective on practical reasoning; again, there is disagreement on whether constructivist practical reasoning does or should employ ideal-theoretical conceptions of moral reasoning or whether it should seek to address ‘real-world’ agents as we find them ‘here and now’.

I am not a constructivist. With regard to Kant, my chief difficulty concerns constructivism’s explicit avoidance of Kant’s practical metaphysics.Footnote ² By Kant’s practical metaphysics, I shall here understand a particular kind of insight that arises from first personal reflection on ordinary moral reasoning.Footnote ³ Kant formally articulates this insight by appeal to his three regulative ideas of reason – freedom, God and the immortality of the soul. His overall claim is that ordinary moral practice would not be possible for us (or at least, would not be sustainable for us) absent practical commitment to these transcendent albeit theoretically indemonstrable ideas.Footnote ⁴ However, it is not Kant’s ideas of reason in particular which here interest me so much as the reflexive form of Kantian moral insight in general. Rawls complains that Kant begins from a metaphysical conception of the person: according to Rawls, this starting point leaves Kant’s moral philosophy mired in foundationalist claims. This is not how I would characterize Kant’s practical metaphysics. Kant does not begin from metaphysics – he rather ends with certain reflexive claims about morality’s necessarily transcendent (hence indemonstrable) status for us; claims that arise from reflection on ordinary moral experience itself.Footnote ⁵ Insight into the transcendent status of morality’s grounds puts us at some distance from our ordinary moral reasoning, inquiry into which was our initial starting point: we observe our morally striving phenomenal selves from a critical vantage point – this has a corrective function, in turn, on what we understand ourselves to be doing when we reason morally.Footnote ⁶

Constructivism is largely uninterested in this reflexive dimension of Kant’s moral philosophy – its focus is on the derivation of shareable substantive principles (of justice or of moral action more generally) for persons conceived as citizens, or as rational agents, or as persons who co-inhabit the world.Footnote ⁷ One might indeed take the view that everyday morality is one thing and reflection upon it quite another, such that what I here call Kant’s practical metaphysics can, without cost, be lopped off from what one might call the normative parts of his theory.Footnote ⁸ However, I believe that, for Kant, reflection on ordinary practical reasoning is integral to being a moral agent.Footnote ⁹ More specifically, insight into the form of practical reasoning provides a practically necessary corrective to ordinary, i.e. substantive, moral reasoning.

The form/content distinction is a familiar Kantian distinction – indeed, it plays a significant role in at least some variants of Kantian constructivism. There, ‘form’ is typically understood in terms of a kind of ‘decision procedure’ that can secure agreement in substance among otherwise differently minded deliberators. My alternative claim here will be that appreciation of the form of moral willing diminishes the significance of abiding substantive disagreement. More mundanely, I shall argue that the shared form of moral willing can accommodate contextually divergent moral norms. I call this Kantian contextualism, meaning by it the idea that the same form of moral willing can legitimately give rise to contextually different substantive principles.Footnote ¹⁰

To some extent, Kantian contextualism grows out of Rawlsian constructivism. Rawls famously holds that his own version of constructivism does not go ‘all the way down’ – a claim that has displeased at least some Kantian constructivists.Footnote ¹¹ Yet for Rawls, the unconstructed building blocks of political constructivism are latently held, socially shared beliefs, the grounds of which Rawls does not enquire into – he simply accepts them as historically given.Footnote ¹² Rawls’s context-dependence has been decried as distinctly non-Kantian by his Kantian critics – it has also encouraged the emergence of so-called Humean constructivism, which has in turn undermined the view of constructivism as distinctly Kantian (Street Reference Street2013). In part inspired by Humean constructivism, David Velleman has recently taken contextualism one step further, towards moral relativism. Significantly, Velleman defends the practical intelligibility of context-dependent norm divergence by way of an appeal to reasons’ action-guiding character. This is why he can plausibly speak of his approach as ‘kinda Kantian’ (Velleman 2009: 149).

I am drawn to Velleman’s Kantian moral relativism in a way in which I have never been drawn to Rawls’s constructivism. This may seem odd, given that Velleman’s position is a development at least in part out of some version (or versions) of constructivism. What sets Velleman’s approach apart from the latter, and what renders it to my mind more genuinely Kantian, is his focus on the idea that reason itself can be action guiding. As we shall see below, Velleman actually speaks of reasons for action rather than of reason; this difference is not insignificant and I shall return to it.Footnote ¹³ Nonetheless, the focus on reasons’ action-guiding character seems to me to closer in spirit to Groundwork than constructivism’s preoccupation with the derivation of substantive principles that are shareable by all reasonable persons. What here interests me more specifically is Velleman’s surprising but plausible conclusion that a focus on agents’ reasons for action will tend to support, not eliminate, norm divergence across culturally different contexts: Velleman’s analysis points away from constructivist aspirations for shareable substantive reasons and towards a Kantian justification of normative differences. I find this to be a welcome development.

Several aspects of Velleman’s account nonetheless also remain ‘kinda Humean’. Velleman shares with constructivists an aversion to Kant’s practical metaphysics; he avoids reference to the noumenal, including reference to agents’ possible reflexive insight into their reasons’ context-dependence. In consequence, while Velleman can show that contextual norm divergence is rationally intelligible, it is not clear that he can show such norm divergence to be morally acceptable from the vantage point of the agents who confront one another with diverging norms. I shall argue that Kantian reflexivity in relation to the form of practical reasoning can help render contextual norm divergence morally acceptable to agents who find themselves confronted by this fact. In what follows, I begin with a summary outline of Velleman’s Kantian defence of moral relativism (section 2); I then go on to identify those Humean elements in Velleman’s account that obscures Kantian reflexivity (section 2). Once these obstructions are removed, the importance of Kantian reflexivity to the moral acceptability of relativism comes into view (section 4). I shall argue that insofar as it is precisely the claim to the reasoned authority of one’s contextual norms that gives rise to inter-communal moral conflict, the resolution of such conflict depends on recognizing the form of willing as morally authoritative over its substantive content (section 5). I conclude with some brief comparative remarks on Rawlsian constructivism (section 6).Footnote ¹⁴

2. Velleman on Practical Reason

In Foundations for Moral Relativism Velleman exhorts his readers to take the social and historical fact of moral relativism philosophically seriously:

Velleman’s focus on moral relativism as a social fact circumvents more standard philosophical disputes about its theoretical coherence;Footnote ¹⁵ he nonetheless acknowledges that even practically minded relativists confront a philosophical challenge: they must show not merely that mores intelligibly differ across cultures but that norms do so as well. Mores are simply culturally specific practices – everyone knows that they differ from context to context. Norms, by contrast, carry the burden of reasoned justification. While different mores may be indicative of different moralities, the challenge for moral relativists lies in ‘bridging the fundamental difference between mores and moral norms’ (Velleman Reference Velleman2013: 47). Velleman proposes to bridge that difference through an analysis of reasons for action: he claims that what transforms mores into norms is the action-guiding authority mores have as reasons for action. Velleman’s basic argument in support of this claim can be summed up as follows: if norms, as reasons for action, are to be action guiding, they must be sensitive to relevant action-contexts. Yet action-contexts differ. Therefore, reasons for action can intelligibly differ from context to context. Reasons’ action-guiding character is then a foundation for – it makes sense of – moral relativism conceived as contextually dependent norm divergence.

In developing this line of argument, Velleman’s first step is to draw attention to the ineliminably first-personal and indexical form of reasons for action. By ‘indexical’, I understand Velleman to mean more than merely ‘first-personal’. First-personal practical reasoning is reasoning from the perspective of the (singular or plural) first person. First-personal reasoning can include reasoning about others’ reasons for action: I can talk about the reasons for action I believe you to have. Indexical reasoning, by contrast, is self-referential first-personal reasoning: it concerns my (our) thinking about the reasons for action which I (we) take myself (ourselves) to have. According to Velleman, to say, on the indexical account, ‘x is wrong’, is implicitly to say, ‘x is wrong for me’. Strictly speaking, the first formulation – ‘x is wrong’ – is merely descriptive; only the second formulation is properly action-guiding in that it expresses my acknowledgement of the proposition’s normative authority for me. Velleman goes on to note that the elliptical formulation, ‘x is wrong’, is nonetheless essential to the proposition’s normative authority for me. Thus, for example, ‘“female circumcision is wrong”, said by a Mbuti is action-guiding; “female circumcision is wrong for the Mbuti” is not’ (Velleman Reference Velleman2013: 47). So although reasons for action are ineliminably indexical, they must have the form of general validity for those whose actions they guide. The elliptical formulation indicates that ‘morality obligates its subjects by being rationally binding on them’ (Velleman Reference Velleman2013: 49). I take this to mean that rational bindingness for me is a function of my judging a given norm to be valid for anyone in my situation. It does not follow from this that everyone in my situation in fact has reason to x: my judgement to that effect merely confirms the rational bindingness of x-ing for me.

Velleman next goes on to say that ‘the relativist must characterize a single relation that reasons always bear to what they are reasons for. His relativism must consist in the claim that one and the same relation is sensitive to differences among communities’ (Velleman Reference Velleman2013: 49). The thought here is that while there may be norm divergence across cultures, the appeal to reasons’ action-guiding character is culturally invariant. But how should one characterize that culturally invariant ‘single relation’ between reasons and actions? Velleman introduces the idea of a weight-bearing relation, which he explicates by way of an analogy: rocks are heavy. In fact, that proposition too, though non-normative, is implicitly indexical: rocks are heavy on earth. Rocks are not heavy simpliciter – they lack gravitational pull in space. Action-guiding reasons are like rocks; whether or not they have directive authority (gravitational force) depends on their context: ‘like a rock, a reason would exert its weight within a frame of reference established by some weight-conferring force’ (Velleman Reference Velleman2013: 52). The single relation that reasons always bear to what they are reasons for is then the gravitational force which they exert on agents within a given frame of reference. This single relation is reason’s ‘constant nature’ – in that sense, there are no different, local methods of reasoning, there merely are different contexts. Whether or not a given custom has the gravitational force of a reason depends on the context: just as rocks have gravitational force on earth but not in space, so some reasons for action have gravitational force in some contexts but not in others.

I am not sure about the accuracy of Velleman’s analogy. Are reasons like rocks or are agents? In Velleman’s analogy, reasons seem more akin to the gravitational force that exerts its influence on rocks within framework ‘earth’. It is then agents who are like rocks when they are pulled by reasons within a given frame of reference. If agents are like rocks on whom reasons exert gravitational force in some contexts and not in others, agents do not give themselves reasons. Instead, they are subject to reasons’ gravitational pull on them: this is a point I shall return to below. For now, let us consider what it is about agents that renders them subject to reasons’ gravitational pull. Here Velleman introduces the idea of a drive towards sociality and a related desire to render ourselves interpretable to each other. We render ourselves interpretable to each other by establishing a common frame of reference that enables us to interpret one another’s motives and actions: ‘the normative force of reasons is the force of the drive towards mutual interpretability, which arises out of the drive toward sociality’ (Velleman Reference Velleman2013: 58; see also Velleman 2009). It is the quest for mutual interpretability that accounts for the emergence of community-specific norms that guide and regulate action and interaction between community members. Human agents are rational animals whose drive towards sociality leads them to seek to render themselves interpretable to one another. Since mutual interpretability depends on shared reasons, human communities establish context-specific sets of norms – frames of reference – that enable them to achieve mutual intelligibility. Moral relativism is the rationally explicable outcome of this general if contextually satisfied human drive towards sociality.

3. The Limits of Velleman’s Kantianism

Two closely related features of Velleman’s account strike me as characteristically Kantian. One is the idea of morality as rationally binding on agents; the other is the indexical nature of Kantian practical reasoning. The first of these features is widely accepted as Kantian – perhaps even as uniquely Kantian – so I will here take it for granted. The second feature is less widely so recognized. The neglect of Kantian indexicality may be a consequence of the widespread constructivist association of Groundwork with the derivation of generally shareable substantive principles (for most Kantian constructivists, the derivation proceeds via the (singular or plural) first-personal perspective).Footnote ¹⁶ At least since Rawls, the categorical imperative test or procedure is widely assumed to be in the service of identifying a set of generally shareable substantive principles of action.Footnote ¹⁷ The basic thought is that any proposed substantive principle, when subjected to the universalizability test, either will or will not turn out to be reasonably acceptable to all. It is worth pointing out that this reading cannot in fact show those principles to result from the categorical imperative procedure – the procedure at best certifies the rational non-rejectability of independently proposed principles.Footnote ¹⁸ In that sense, the Rawlsian reading tends in fact to substantiate the Hegelian objection that no amount of rational form will of itself generate substantive content.Footnote ¹⁹

More pertinent in the present context is Rawls’s neglect of the central role of agents’ maxims in Kant’s own account of the categorical imperative. Groundwork is not in fact concerned with the derivation of generally shareable principles of morality from some specified formal procedure. The text’s chief concern is instead with the idea and practical possibility of unconditionally good willing.Footnote ²⁰ Right away, this places the focus on the individual agent – more specifically, on the agent’s maxim, which emanates as a potential reason for action from her will (the latter is itself conceived by Kant as a ‘kind of causality’ capable of effecting action). An agent’s maxim is accessible only to herself – a maxim is a subjective and indeed ‘inner’ volitional principle (cf. G, 4: 421). To the extent then, that the categorical imperative serves as a test for the agent’s proposed maxim, that test is necessarily indexical in structure: the agent asks herself whether in her judgement her proposed maxim is serviceable as a possible principle of action for everyone. The point of the universalizability test is not to generate universally valid principles of action but to judge the moral purity of one’s maxim: the universalizability requirement constrains a finite rational agent’s natural tendency to judge in her own favour by asking her to consider whether, in her judgement, her proposed maxim could qualify as universal law. It does not follow from this that the agent’s judgement in the matter is authoritative for everyone else; indeed, it is hard to see how it could be given that each has legislative competence only with regard to her own maxims.

On the reading here sketched, the categorical imperative is, first, not about the generation of shareable substantive principles but about the reflexive evaluation of the moral form of one’s own proposed maxim. Second, only the agent herself has insight into her maxim, so only the agent herself can judge its moral form.Footnote ²¹ Third, the agent’s judgement of universal validity can hold valid only for the agent herself: she cannot legislate the moral law to anyone other than herself since she has neither insight into anyone else’s maxims nor the moral authority to judge them. Kant thus agrees with Velleman that reasons for action are ineliminably indexical. While in judging a given maxim morally permissible for me I do raise a universality claim, that claim is structurally analogous to Velleman’s point about the implicitly indexical nature of reasons for action. Recall: according to Velleman, in accepting a given reason for action as valid for me I accept it as valid simpliciter – however, my judgement to that effect merely confirms the rational authority of x-ing for me. Analogously, in judging my proposed maxim universalizable, I judge it to be rationally binding on me – universalizability is the form of reasoned authority in general.

Despite their shared emphasis on the indexical character of reasons for action, there is also an important difference between Velleman and Kant.Footnote ²² It concerns the grounding relation between reasons and action – what Velleman calls the ‘gravitational pull’ of reasons. We saw that, although Velleman likens reasons to rocks, the better analogy may be between rocks and agents. Velleman’s agents are similarly as passive in relation to reasons as rocks are in relation to earth’s gravitational force: neither is pulling themselves by their own volition. This initial impression of agential passivity is confirmed in Velleman’s subsequent appeal to a drive towards sociality. We do not act on reasons through our own volition but in response to the drive towards sociality. The overall picture that emerges from initial Kantian beginnings is of a broadly Humean, naturalized conception of practical reasoning that is in the service of some other, independently specified end. For Kant, by contrast, we do act morally from our own volition; moreover, we so act for its own sake. Kant’s emphasis on agents’ own judgement in relation to their proposed maxims attests to the volitional nature of moral agency for him: agents act on those maxims which they judge to be morally permissible (or required) and they act on them because they judge them to be morally permissible. (G, 4: 451–2; CPrR, 5: 43).

It is in respect of our volitional moral capacity as the determining ground of moral agency that Kant introduces the idea of our part noumenal status. Our volitional moral capacity as the ground of our ordinary moral agency comes as a surprise to us – unexamined ordinary moral reasoning suggests multiple, rather different grounds (and ends) of morality (G, 4: 294–6). We had not suspected ourselves capable of moral volition in the strong sense suggested by Kant’s analysis of ordinary moral reasoning. Nor can we ultimately specify the ground of that capacity in turn – we can only ‘comprehend the incomprehensibility’, for us, of those grounds (G, 4: 463). To that extent our reflexive insight into our own volitional powers remains problematic for us: we must take ourselves to have volitional moral powers whose ultimate source we must also acknowledge as unfathomable for us.

Although Kant’s introduction of a noumenal perspective on our ordinary moral reasoning is both problematic and controversial, it is integral to his moral thinking: without it, we could not gain reflexive insight into the form of ordinary moral reasoning.Footnote ²³ I shall here neither defend nor reject Kant’s claims in behalf of our part-noumenal status but shall simply accept them at face value. Of interest in the present context is the thought that, in taking the noumenal perspective upon ordinary moral reasoning, we do in a sense transcend the latter’s contextual constraints. In Groundwork, which is focused on inner freedom or morality (virtue), these constraints are inner ones – subjective desires and inclinations: the critical perspective enables us to acknowledge that we can act independently of the constraints of personal happiness, say, or comparative advantage. Outer constraints, such as the habitual influence of culturally specific mores on moral judgement do not surface in Groundwork; they are the subject, rather, of Kant’s political writings, and of his cosmopolitan writings in particular.Footnote ²⁴ Even so, the distancing effect in relation to ordinary moral reasoning is plain: in taking the noumenal perspective, the agent acknowledges that it is the form of willing – good will – that matters morally, not its particular substantive content.

Velleman sidesteps the noumenal aspects of Kantian morality. He takes morality to be a much more mundane enterprise: ‘morality is distributed holographically throughout our lives, in the various ways that our shared practices and values reflect various rational pressures and the underlying human nature in light of which those pressures have been accommodated’ (Velleman 2009: 154). Indeed, Velleman may regard the ordinariness of morality as a function of its reason-giving nature: if to be action-guiding, reasons must be sensitive to context, then it may be endangering reason’s action-guidingness to entertain ideas about context-transcendence. That is a plausible objection to Kantian noumenalism so far as it goes; however, it comes at a price. As noted, it remains crucially unclear on Velleman’s account to what extent it is in fact reason that guides action. We engage in mutual reasoning in order to render ourselves intelligible to each other. Yet it is not clear how much of a grip we have on this drive towards sociality. We seem rather to be in its grip. We throw each other socially constructed cues for interacting, which we interpret as we go along. We are aware of the fact that we are thrown cues by others to which we respond and are aware that we in turn throw others cues to which they respond. We are aware that they are aware that we are aware. And so on. Despite all this mutual awareness of each other, Velleman’s agents seem to remain reflexively unaware of what each is himself doing. The form of their moral reasoning never surfaces into their consciousness – Velleman’s agents remain wholly unaware of the fact that ‘the same single relation between reason and action’ characterizes the basic structure of all substantively divergent moralities. For Kant, by contrast, the unconditional nature of morality does impinge on our consciousness, pulling us up short and occasioning a suspension of our unreflective assumptions about ordinary, substantive moral reasoning.Footnote ²⁵ Groundwork is designed to leave us hyperaware of the reflexively established insight that, in matters moral, it is the form of willing that matters, not its substantive content. In what follows I shall suggest that the moral acceptability of moral relativism may depend on agents’ reflexive ability to acknowledge invariance in form, even despite variation in substance.

4. The Shock of Moral Relativism

What is problematic for us about the fact of moral relativism is not that other people do things differently, rather it is what that fact about them implies with regard to our own moral commitments. Mundanely, the fact that other people do things differently shows us that ours is not the only way in which to do things. That is fine where only mores are concerned; it becomes undermining of deeply held convictions where morality is at stake. Morality is supposed to have universal validity – certainly a morality built on the authority of reason. The appeal of Foundations for Moral Relativism lies in the way in which Velleman mobilizes the Kantian idea of reasons for action against claims on behalf of a substantive universalism: he shows that it may be precisely the idea of a reason for action that can render the fact of norm divergence rationally intelligible to us. And yet Velleman seems in the end to succeed only by coming away with a severely deflated conception of morality. While norms are shown to be distinct from mores – they are shown to be reasons for action – this may turn out to be a distinction without a difference. For if ‘morality is distributed holographically throughout our lives, in the various ways that our shared practices and values reflect various rational pressures and the underlying human nature in light of which those pressures have been accommodated’ (Velleman 2009: 154), then frankly we could have spared ourselves the trouble of seeking to differentiate norms from mores. Norms are pragmatically justified mores – justified with reference to how well they enable us to get along. Norms might as well simply be mores.

To see just how deflationary Velleman’s account of morality turns out to be, let’s consider what happens when two communities encounter one another – each from within its own particular frame of reference. Oddly, on Velleman’s account, nothing much happens. There is no relativistic stand-off – no squaring up of one community’s norms to those of the other. Instead, Velleman simply assumes that the human drive for sociality will extend outward. Members of each community will seek to render themselves interpretable to members of the encountered community. The ensuing game of mutual cue-giving across communal borders ensures that a new inter-communal frame of reference is gradually established. One might think of trade relations as emerging in this way. Each party is curious about the other, the curiosity may be motivated by ulterior concerns, yet in the course of pursuing that concern the trading parties render themselves intelligible to each other at least to the extent necessary to facilitate mutual trade. In time, these new inter-communal norms may affect each set of intra-communal norms so that eventually, all three may gradually converge into a single, enlarged frame of reference.Footnote ²⁶

Velleman does acknowledge that some of the norms practised by one community may elicit strong disapproval by members of the other community. On his account, these norms cannot be dismissed as morally aberrant by members of the objecting community: they are bona fide reasons for action for the community whose norms they are. Yet outsiders who abhor those practices can account for their prevalence in those communities in ways that avoid direct moral confrontation: they might treat the prevalence of these practices as indicative of the community’s backwardness, for example. It is then not the case that, relative to their contexts, these practices do not constitute reasons for action. Rather, something else about these contexts may be amiss which we might want to change in order to encourage a change in reasons for action. Velleman sensibly warns that, even so, ‘reason-guided change is path-dependent: where it ends up largely depends on where it began. Different communities may have reason to change in ways that still lead to different ways of life’ (Velleman Reference Velleman2013: 68). We cannot compel others to have reasons for action which their contexts do not make available to them: ‘the rational way to disagree with those who live differently is to articulate our own self-understanding, listen as they articulate theirs, and then go back to our respective experiments to see whether we have learned something by which to understand ourselves better by living differently’ (Velleman Reference Velleman2013: 69).

I am sympathetic to the idea that the best way in which to come to understand each other across the cultural divide is by talking and listening to each other. At the same time, Velleman’s intimation that this is the typical way in which culturally differentiated people in fact do encounter one another depends on his holding a conception of morality that is so undemanding as to make it a mystery how the practical problem of moral relativism could ever arise in the first place. We encounter each other, observe our respectively different ways of going on, either do or do not learn from each other, and then each go on our merry way again. This is not what encounters of moral difference have looked like historically. Velleman’s argument is highly plausible, that the fact of moral relativism can intelligibly be understood as arising out of reasons’ action-guiding character. But in a sense, this simply ups the ante, for there is then a tension internal to the structure of reasoned morality between the ineliminably indexical character of reasons for action and rationality’s equally authoritative claim to universal validity. It is because we believe our moral norms to be rationally justified that we react with hostility to other cultures’ different moral norms.

Let me suggest an alternative phenomenology of moral difference to the one implied by Velleman’s account: the fact that others do things differently from us undermines our confidence in our own norms precisely because, having the authority of reason on our side, we treat our norms as universally valid. Others’ norms and values then pose a threat to our confidence in the rational validity of our own norms. Our first reaction is likely to be one of protectiveness towards our own norms, not curiosity about others’ norms. Our second reaction is likely to be hostility towards (the norms of) the other community. Our third likely reaction is to attempt to persuade members of the encountered community of the errors of their moral ways – by force if necessary.Footnote ²⁷ Much of this will be done in the service of our maintaining confidence in our own moral norms. The question is whether we can step back from that non-reflexive chain of reactions in order to ask ourselves whether that which we naively took to be the moral significant about our set of substantive norms really is so at all.

5. Kantian Contextualism

I believe that my alternative phenomenology of cross-cultural encounters has the merit of greater historical accuracy over Velleman’s more peaceable, intrinsically more attractive account. My alternative phenomenology is also designed to bring out why a morality that purports to be grounded in reason is likely to elicit particularly strong claims on behalf of its universal validity. If reasons guide action, then given that reason is the ultimate court of justificatory appeal available to us, the natural assumption on each side of a communal encounter must be that it is in the right about its moral norms and the other side is in the wrong about theirs. Velleman’s analysis targets that natural assumption: if reasons do guide actions, then different reasons will in fact be called for in different action-contexts. It turns out that reasons’ action-guiding character in fact supports substantive moral relativism rather than substantive moral universalism. As noted, I find this quite compelling; the problem is that Velleman’s defence of moral relativism is not internal to the conception of moral reasoning he attributes to his moral subjects. While Velleman is cognizant of the context-dependent character of reasons for action, his moral agents appear not to be; they lack reflexive insight into that ‘single relation’ between reason and action which turns mere mores into norms for them. Lacking that insight, they seem to me more liable to react with hostility than curiosity to others’ culturally alien norms: they lack the resources to acknowledge that the same single relation between reason and action may be responsible for substantive norm divergence.

I said earlier that Kant’s noumenal standpoint affords us a perspective upon ordinary moral reasoning that both arises from inquiry into the activity of ordinary moral reasoning and transcends it. In Groundwork, this noumenal perspective is articulated with reference to our comprehension of the incomprehensibility, for us, of the ground of our practical freedom. In the second Critique the point is put in terms of practical freedom as a ‘fact of reason’.Footnote ²⁸ Both times the immediate subject matter is our reflexive awareness of our (theoretically indemonstrable) practical freedom. I suggested, however, that the Kantian insight can be extended towards a more immanent, Velleman-like way of thinking: reflexive awareness of ourselves as practically free is reflexive awareness of the underlying volitional structure of our ordinary substantive moral reasoning. For Kant, what Velleman calls the ‘single relation’ that underlies all substantive moralities is the agent’s endeavour to align her subjective maxim with what she judges to be its universally valid version. The Kantian analogue to Velleman’s single relation is the reflexively judged universalizability of individually held maxims. And that feature of human morality may be invariant across normatively divergent contexts for Kant in the same way in which the single relation between reasons and action is invariant across normatively divergent contexts for Velleman. The crucial difference between Kant and Velleman is that it matters to Kant that agents be reflexively aware of the form of their willing as that which makes the moral difference – i.e. accounts for the moral worth of their actions. It is precisely in this sense that Kant claims a critical function for Groundwork: while the analysis tracks ordinary moral reasoning, Kant’s focus on the form of moral volition at the same time lays bare what makes ordinary morality moral in the first place.

Admittedly, neither Groundwork nor the Critique of Practical Reason so much as raises the spectre of moral relativism. To the contrary, Kant assumes that the substantive deliverances of agents’ categorical imperative tests generally will dovetail with the principles and precepts of ordinary moral reasoning. Kant assumes that any competent moral reasoner will judge it wrong arbitrarily to break promises, will judge it wrong not to help others when in a position to do so, will judge it wrong to commit suicide in order to escape the burdens of life, etc. Indeed, Kant appears to assume that attention to the form of moral willing vindicates ordinary substantive judgements. He does not appear to treat the rule against promise breaking as a context-dependent reason for action but seems to see it, rather, as a universally valid deliverance emanating from his proposed focus on form. That, at any rate, is how Kant’s practical philosophy is generally read.

But consider this. Of any one agent who asks himself whether he should keep a promise to repay a sum loaned or should default on repayment so as to help a third party in need, which way would Kant have this agent jump, ethically speaking?Footnote ²⁹ I am inclined to say that Kant’s argument in Groundwork constrains him to let the agent judge which of these two possible courses of action contains the greater ground of obligation for her. If the agent sincerely judges either maxim universalizable – repaying the loan or defaulting on repayment to help someone in need – she has done all that Kantian morality can reasonably demand of her. Admittedly, the particular example trades on the idea of a possible conflict of duties: in such cases it may seem more intuitive to let the agent judge for herself. Yet it is well known that Kant denies the possibility of a conflict of duties: the moral law cannot pull you in two directions at once. You must yourself judge, therefore, the morally requisite ground of will-determination for you.Footnote ³⁰ I believe, however, that the point generalizes: if a reason for action has to be a reason for me, then it has to be a reason that stems from my judgement to that effect. At least in the domain of ethics – which is the concern of Groundwork – our judgements are necessarily each our own. I cannot judge the purity of your maxim and you cannot judge the purity of mine. Who is to say, then, that of two different agents in structurally similar circumstances that one may not sincerely judge repayment of the loan to be morally requisite, while the other equally sincerely defaults on it to help a third party?

Its focus on maxims and purity of will point to the radically subjective nature of Kantian moral judgements. This is not to say that these judgements are subjectivist – that they are based on particular interests, preferences or inclinations. To the contrary, the subjectivity of Kantian moral judgement and will formation should be read in the spirit of Velleman-type indexicality: Kantian moral judgements are subjective in the sense of being self-addressive: they are judgements concerning valid reasons for action for me, where the criterion of validity for me is universalizability of maxim. My judging my maxim universalizable in principle is my judging it rationally authoritative for me. So long as the agent sincerely judges her maxim to be universalizable, substantive divergence from others’ formally identical such judgements is morally irrelevant.Footnote ³¹

True, the subjective nature of Kantian moral judgement does not to equate to moral relativism or contextualism as standardly conceived. Both of the latter refer to cross-communal differences; neither standardly entertains the possibility of community-internal divergence in subjective judgements. The reading of Groundwork here sketched implies, one might say, a kind of hyper-relativism regarding community-internal individual judgements! I think this is correct in principle, though it is worth noting that Velleman’s account would imply the same if he were to replace what he calls the common drive towards sociality with a more strictly volitional account of indexical reasons for action. Either way and conceding that the subjective nature of Kantian moral judgement is distinct from moral relativism as standardly conceived, there is at least a route from Kantian subjective moral judgement to Kantian friendliness towards context-dependent norm divergence. If it is the form of willing that matters morally across substantively diverging community-internal individual judgements, then it must also be the form of willing that matters when it comes to substantively divergent cross-communal individual moral judgements.

In ‘Maxims and Thick Ethical Concepts’, Adrian Moore reaches similar conclusions, albeit from a focus on the role of ethical concepts in relation to maxim formation rather than from a focus on the form of Kantian moral willing (Moore Reference Moore2006). Moore’s analysis has the added virtue of showing why intra-communal norm convergence is likely even despite the essentially subjective (i.e. indexical) nature of Kantian moral judgements: according to Moore, Kantian maxims are necessarily context-dependent given that their formulation involves thick ethical concepts. A thick ethical concept is always responsive to a socially established practice – practical concepts are not sui generis. Take the concept of promising: in order sensibly to ask myself whether a maxim of breaking a promise whenever it is convenient for me to do so is universalizable I must be familiar with the practice of promising – I cannot have a maxim of promise keeping unless I know the meaning of promising. Similarly with formulating maxims that involve concepts such as ‘lying’, ‘helping others in need’, ‘committing suicide’, etc. Yet to the extent to which my formulating maxims is context-dependent – i.e. to the extent to which it involves appeal to thick ethical concepts – the substantive conclusions to my formal universalizability judgements will be context-dependent also. One may object that at least some moral concepts are, if not sui generis, at least widely shared across culturally different communities: concepts such as promising, for example, or lying or helping others in need. Concepts such as these, one might object, are as good as being moral universals. But it is then better to say, with Velleman, that these concepts are ubiquitous, that we find versions of them in most and perhaps even in all human cultures. Even so, the inference from the centrality of some of our thick ethical concepts for us to their universal validity is unwarranted; Velleman, for one, offers a truly astounding list of examples that attest cultural divergence from norms which we tend to regard as universally valid from our own culturally parochial perspective (Velleman Reference Velleman2013: 23–44). Your culture may lack the practice of promising – certainly many cultures appear to lack the practice of contracting: if so, formulating maxims of reliable promise keeping is morally unintelligible. Similarly, your culture may treat as morally significant practices which my culture regards as of peripheral moral importance at best: filial loyalty, for example. The maxim of putting filial loyalty above personal fulfilment may be a moral fundamental in your culture but not in mine. In formulating personal maxims agents cannot but draw on contextually established practices and relevant thick ethical concepts responsive to those practices; in subjecting these maxims to the universalizability test, they will necessarily come away with judgements that are context-dependent substantively speaking. From Moore’s perspective, then, Kant’s confidence that his critical analysis of moral willing leaves ordinary substantive morality unaffected shows that Kant labours under context-dependent norms when he takes it as self-evident that we will judge suicide morally wrong, etc. By the same token, the structure of moral willing remains unaffected by substantive dependence on thick ethical concepts.

Is this not a pyrrhic victory, however: are we not condemned to move in a vicious circle when our critical testing of our maxims is itself dependent on uncritical acquiescence in context-dependent ethical concepts at the level of our formulating our maxims? I do not think so: judging the moral probity of our maxims may well include critical reflection on the practices and concepts that enter into maxim formulation; thick ethical concepts may change in response to our moral evaluation of maxims whose formulations unavoidably draw on those concepts. Of course, one should not overestimate the power of individual moral judgement in this regard: moral critique and social change depends on many things; it is generally gradual and of an order of social magnitude that exceeds individual will formation – Kant’s focus in Groundwork. Certainly critical reflection on our communal norms and thick ethical concepts depends on a willingness to engage with those who do not necessarily share those norms and concepts – it depends, as Velleman says, on a willingness to talk across community boundaries. What Kant’s analysis in Groundwork does offer us in this respect is the reflexively grounded ability to predicate a morally good will even of those who do not share our substantive moral norms. To the extent to which we are critically aware of the necessary context-dependence of our own maxims as reasons for action for us we might be able to appreciate the context-dependence of others’ maxims as reasons for action for them. So long as others sincerely judge their maxims to be universalizable from within the contexts they find themselves in, it will be hard for us to find them morally wanting by the standards of moral judgement we apply to ourselves.

6. Constructivism, Contextualism and Moral Relativism in Kant

Should we read Kant contextually? Many will think that it matters that we not read him contextually but that we instead hold the Kantian line on behalf of moral universalism. I believe, however, that nothing that I have here said tells against Kantian moral universalism – the argument here developed only tells against a universalism of substantive moral norms and principles. I believe that while the categorical imperative is meant to give us the form of moral judgement in general, this does not mean that it can tell us which substantive moral principles are valid for or true of all people at all times. To the contrary, if it is judgement that counts, then truth is out of the running. If there were a truth about morality, we would not need to judge what we ought to do – we would rather need to find out what that truth consists of. But just as we do not judge that 2+2 = 4, so we do not know what we ought to do in any given situation: rather, we reach a judgement to that effect. In reaching that judgement the best we can hope for, at least according to Kant, is that we judge and act from purity of will – which I here take to mean that we judge and act to the best of our ability, morally speaking. This does not mean that we can never get it wrong – if anything, Kant seems to believe that we are more liable to get things wrong than right, meaning that we more often fail to act from purity of will than succeed at so doing. Still, there is nothing over and above the act of judgement that can tell us, either after the fact or before it, that what we proposed to do or ended up doing was either the right or the wrong thing to do. Nor does this mean that there cannot be moral disagreement. I can judge you to be doing the wrong (or the right) thing and you can judge me to be doing the wrong (or the right) thing. We can query the purity of one another’s maxims from the outside, as it were: but we cannot know whether the other’s maxim truly was pure; indeed, Kant denies such knowledge even with regard to the introspective inspection of our own maxims. Kantian ethics is subjective (i.e. first-personal indexical) judgement all the way down.

Most readers of Kant seem to be looking for more, however. While many agree that ‘Kant thinks that no one else can judge on our behalf what we ought to do’,Footnote ³² they also insist that there are determinate rights and wrongs and that each particular reasoner’s application of the categorical imperative somehow tracks those rights and wrongs. Some suggest that correct application of the categorical imperative test will generate the right outcome; they seem to think of relevant standards of correctness observing basic rules of logic or of correct reasoning more generally. But while necessary, rules of right reasoning cannot in themselves deliver substantive moral judgements. Others think of the categorical imperative test more along Rawlsian lines, i.e. as a decision-making procedure that will yield convergence in judgement among reasonable persons. Yet it is not clear in what sense the assurance that others agree with my judgement can help me determine the purity of my maxim: convergence in judgement seems to be the wrong criterion for the task at hand.

My basic claim is that those who look for a universalism in moral substance overlook the universalism in the form of willing in Kant’s ethics. Granted, an emphasis on universal form over substance is itself a far cry from relativism – it turns out that we do share something across individuals and cultures alike. And perhaps that fact – the fact that we share the form of moral willing – is enough to get meaningful substantive moral disagreement going. Of course, to be moral, the disagreement has itself to be sincere – i.e. it has to reflect disagreement over moral norms and principles, not other considerations, such as those of power and politics. Where moral disagreement is sincere, i.e. where we discover that others differ from us in their moral judgements as to what it is that they (and by extension, we) ought to do, such disagreement can, as Velleman says, be a powerful lesson in humility.Footnote ³³

I want in conclusion to briefly return to Rawls. I said that, for Rawls, constructivism does not go all the way down; he relies on certain building blocks, specifically, on socially transmitted commitment to the values of freedom and equality. These form the basis of his ideal-theoretical construction of the person and correspondingly attainable well-ordered society. Rawls’s constructivism is thus relative to a particular value scheme which is itself simply set out. Some of the implications of Rawls’s value-relative constructivism come to the fore in The Law of Peoples. Rawls there denies that justice as fairness applies to non-liberal social contexts; he goes on to claim that the legitimacy of social institutions in what he calls decent hierarchical societies is a function of how well those institutions reflect the relevant social background values operative in those societies. In short, Rawls’s own constructivism endorses value contextualism to a significant extent.

Rawls’s moral concessions towards non-liberal societies have widely been decried as a betrayal of his liberalism and of his Kantian credentials. The interpretation of Groundwork here sketched may to the contrary seem to vindicate the Kantianism of Rawls’s value-relative constructivism. I also said, however, that Rawls’s acceptance of historically given background values follows from his rejection of (what he takes to be) Kant’s practical metaphysics. On my reading, therefore, Rawls embraces value relativism (or contextualism) as a result of his rejection of Kant’s practical metaphysics. Yet Kant’s practical metaphysics is in fact indispensable to the Kantian contextualism sketched in this article. This suggests that Rawlsian contextualism and Kantian contextualism are two distinct kinds of contextualism. In the one case, the contextualism is a function of culturally specific value premises – freedom and equality. In the other case, contextualism is a function of what one might call a subject’s culturally situated exercise of moral judgement. Again, these may seem to come to more or less the same thing: Rawls starts with a determinate set of culturally specific values, whereas Kant’s account of moral judgement allows ‘thick ethical concepts’ into individual maxim formation and testing. I believe, however, that Velleman’s focus on reasons for action can illuminate the basic contrast between Rawls and Kant. The relevant differences are the following: Rawls starts from a socially shared commitment to freedom and equality and asks which scheme of social justice it would be reasonable to agree on relative to those values. The original position – a decision-making device modelled on the categorical imperative ‘procedure’ – yields an outcome, which all who do share those values can agree on as right or reasonable. What makes the outcome reasonable is the fact that all who do share those basic values can accept the eventual outcome as plausible extrapolations from those values. What drives the entire argument are the shared values and what we can get out of them for the purposes of social justice. The idea of a reason for action never really surfaces in Rawls’s account. In Velleman’s account, by contrast, the idea of a reason for action is central. For Velleman, for a given value or norm to be morally authoritative for you it has to be a reason for action for you, and for it to be a reason for action for you, you have to acknowledge it as such. This seems to me to come much closer to Kant’s intention in Groundwork: what the universalizability test asks is not whether my judgement agrees with everyone else’s judgements; rather, the test demands that I ask myself whether I judge my proposed maxim to be a possible maxim for anyone in my situation. Velleman nonetheless goes on to specify the drive towards sociability as the motivating force behind my endeavour to act from reasons. Ultimately, I act from reasons because doing so is conducive to some other end I have: I want to render myself intelligible to others. For Kant, as for Velleman, something is a reason for action for me just in case I judge it so. Again, for Kant as for Velleman, I draw on thick ethical concepts in forming maxims and in subjecting them to the universalizability test. Yet for Kant I do not do so in order to satisfy my drive towards sociality. In a sense, the fact that I draw on the particular thick ethical concepts of my social context is contingent and, therefore, morally irrelevant. What is morally pertinent is my judgement of my thick ethical maxim as a reason for action and my acting on it for that reason. Whilst Rawls is not, in the end, concerned with reasons for action at all, that is all Kant is concerned with. And it is precisely because that is all that Kant is concerned with that Kant’s contextualism is not undermining of his moral universalism.Footnote ³⁴