A Law of One's Own: Self-Legislation and Radical Kantian Constructivism

Authors


Abstract

Radical constructivists appeal to self-legislation in arguing that rational agents are the ultimate sources of normative authority over themselves. I chart the roots of radical constructivism and argue that its two leading Kantian proponents are unable to defend an account of self-legislation as the fundamental source of practical normativity without this legislation collapsing into a fatal arbitrariness. Christine Korsgaard cannot adequately justify the critical resources which agents use to navigate their practical identities. This leaves her account riven between rigorism and voluntarism, such that it will not escape a paradox that arises when self-legislation is unable to appeal to external normative standards. Onora O'Neill anchors self-legislation more firmly to the self-disciplining structures of reason itself. However, she ultimately fails to defend sufficiently unconditional practical norms which could guide legislation. These endemic problems with radical constructivist models of self-legislation prompt a reconstruction of a neglected realist self-legislative tradition which is exemplified by Christian Wolff. In outlining a rationalist and realist account of self-legislation, I argue that it can also make sense of our ability to overcome anomie and deference in practical action. Thus, I claim that we need not make laws but can make them our own.

I

Constructivists have adopted the Kantian motif of self-legislation in arguing that normativity should be understood in terms of rational agency.1 The most militant constructivists claim that all normative authority in practical reasoning is ultimately self-legislated. These constructivists believe that a person can only ever be subject to their own laws. In this article, I explore two leading accounts of self-legislation, whilst arguing that neither can escape endemic problems generated by radical constructivist metaethical commitments. In light of these problems, an alternative realist history of self-legislation is unearthed and examined, which suggests self-legislation can be usefully reconceived as rational appropriation of an already-authoritative law.

What does giving a law to oneself mean? Radical constructivists use the image of self-legislation to illustrate the claim that legitimate authority over us must be authorised by us. The laws to which someone is subject are to be determined by the evaluative commitments they can or do consistently endorse from the practical standpoint of a rational agent deciding how to act. Thus, normative authority in practical reasoning is meant to be an authentic manifestation of whom and what a person is, rather than a fundamentally alien imposition. Yet, at the same time, self-legislation promises to ground principles that can govern and genuinely bind agents, including robust moral obligations. The result is a modernist ethics which many will find attractive: desires, deliberations, and social roles should be accepted as normatively binding upon agents only insofar as they are consonant with human self-determination.

II

The resurgence of self-legislation as a metaethical concept is primarily due to the rise of constructivism in moral and political philosophy. Contemporary constructivism has its roots in John Rawls' reading of Kant's practical philosophy in A Theory of Justice.2 He claims:

Kant's main aim is to justify and deepen Rousseau's idea that liberty is acting in accordance with a law we give to ourselves.3

In subsequent work, Rawls avoids the language of self-legislation, talking more often of citizens regarding themselves as ‘self-originating’ and later ‘self-authenticating sources of valid claims’.4 Nevertheless, autonomous institution or validation of normative authority remained a concern of his.

The Rawlsian articulation of constructivism has been even more significant for reviving the concept of self-legislation than his explicit discussion of it. His political constructivism aims to secure genuine objectivity for the normative judgements which are to guide our actions; and it seeks to do this by appealing to a valid procedure for arriving at these judgements.5 This reverses the standard direction of justification and explanation: justified normative judgements are now those that result from correctly following a correct procedure.6 In contrast to ‘rational intuitionism’, the correct procedures model principles of practical reason rather than mirror an independent moral order. The famous Rawlsian ‘original position'—which instructs us to consider what principles rational agents behind a ‘veil of ignorance’ would choose—qualifies as such a procedure when its results are brought into reflective equilibrium with our considered moral judgements.

Crucially, Rawls’ political constructivism is limited in various respects. He insists that ‘not everything can be constructed and every construction has a basis, certain materials, as it were, from which it begins’.7 We can identify both horizontal and vertical limits to the project. Horizontally, it is only one kind of principle that is constructed: political principles of justice governing the basic structure of society. Other practical or theoretical norms—such as further moral, legal, aesthetic or scientific standards—remain unconstructed. Vertically, Rawls' justifications do not go ‘all the way down’. The correct procedures are those that model the principles of practical reason—which are not themselves constructed—and furthermore they depend upon a ‘Kantian’ conception of persons as free and equal which is itself in need of justification. The later Rawls decisively commits himself to a political and not metaphysical justificatory strategy, which appeals to a conception of personhood implicit in ‘our’ public political culture.8 Ultimately, he attempts to remain neutral about the metaphysics of reason. This is so that a political liberalism based upon a constructivist methodology can be compatible with a wide range of comprehensive doctrines which presuppose other metanormative grounds, such as divine command theories or rational intuitionist forms of realism.9 I shall group under the heading ‘moderate constructivism’ those positions which can be grounded in non-constructivist accounts of normativity in this way.10

III

Among Rawls' students, there were some who thought that constructivism should annex more territory in practical philosophy. We face the rather unlikely image of militant Rawlsian partisans here, who are keen to storm the fortress of practical reason itself. These radical constructivists deny that there must be some basic normative concepts that remain unconstructed: there are to be no lower limits to construction. Furthermore, they push outwards beyond political justice, constructing the normative architecture of the domains of morality and practical reason in general.

Radical constructivists have returned to Kant for the very reason that Rawls departs from him: the promise that fundamental ethical principles can be constructed by autonomous agents.11 Self-legislation is one of the Kantian resources which they use to articulate a defence of the claim that the only intelligible ground for normative authority in practical deliberation is the rational will's binding of itself. Since other sources of normativity cannot be presupposed, the radicalisation of constructivism imposes heavy explanatory and justificatory burdens. Without the more fundamental normative substrata which positions like metaethical realism or expressivism provide, self-legislation must confront a paradox.12

The paradox of self-legislation arises when we ask whether legislating is governed by antecedent norms. If there are no such norms, then self-legislation would be blind—being no more than an ‘arbitrary self-launching'—and so unfit to express of our freedom or ground reasons or values.13 But if self-legislation is governed by antecedent norms, then their status remains in question, since we can ask whether these higher-order norms are self-legislated. The inquiry can be reiterated if these antecedent norms are also self-legislated; and for self-legislation to begin at all, it cannot be that there is an infinite series of antecedently self-legislated norms ‘all the way down’. But if at any stage self-legislation must rest on non-self-legislated norms, radical constructivism must be false as a thesis about all normative authority in practical reasoning.

Two dimensions of the paradox should be distinguished, which can be framed by questions to a potential legislator. First, why should you self-legislate at all?14 If the only source of reasons is self-legislation, then before self-legislation is undertaken, there could be no reason to undertake it. Reasons to legislate which are not somehow self-legislated would demonstrate that radical constructivism is not a comprehensive account of normativity. Second, how should you self-legislate? If there are already reasons that determine what we should legislate before we do so, then the legislative process looks to be redundant and based on non-self-legislated grounds. Whereas, if there are no such reasons, then what we legislate is capricious—being no more than a faux-existentialist salto mortale, which constructivists should want to reject.15 The fundamental dilemma facing the radical constructivist is that self-legislation is either redundant or arbitrary.

IV

Consider two responses to the dilemma. Moderate constructivists are able to accept the self-legislation of normative authorities whilst looking well-placed to avoid both horns. This is because these constructivists can appeal to some other norms to underpin actual or hypothetical deliberation of agents who are to give themselves principles. The vertical limits on moderate constructivism mean that there is no commitment to denying antecedent practical norms. Therefore, unconstructed norms are available to motivate and shape legislation (e.g. those incorporated in Rawls' conception of personhood).16 Since there are reasons why and how legislation should be undertaken, it does not have to proceed randomly, wilfully, or unjustifiably, and so the arbitrariness horn is avoided. Secondly, when self-legislation is embedded within a ‘device of representation’ for the purpose of ‘public reflection and self-clarification’ concerning our existing commitments, then moderate constructivists can avoid the redundancy horn.17 This is because appeal to self-legislation would have a clarificatory role in helping to explain the relation between agency and legitimacy, but without foregrounding the autonomous creation of new norms in any weighty sense which subservience of legislation to anterior normative authorities would render pointless.18

Another response to the dilemma comes from radical ‘Humean constructivism’, whose proponents think self-legislation is the ultimate source of practical norms, yet deny that arbitrariness is a problem.19 For instance, Sharon Street holds that reasons for action arise from the practical standpoint of agents who legislate to themselves, whilst rejecting the Kantian claim that there must be some reasons common to all such standpoints. Instead, practical normative authority is simply a function of judgements a person makes on the basis of their subjective motivational set. Minimal rational constraints arise inter-attitudinally: what an agent legislates constrains what else they can legislate. Antinomies cannot be legislated, for example. Thus, the requirement to both to act and not act in a certain way is disallowed—such as, to always maximise profit and never to extract profit from unsafe working conditions. These formal inter-attitudinal constraints mean that legislation is not completely haphazard. However, Street denies that there are any ‘substantive’ constraints which practical reason or any other authority can provide to govern legislation, and she self-consciously embraces relativism about value.20 Unlike their Kantian cousins, Humean constructivists do not think that self-legislation must be moral or avoid arbitrariness: the law is determined by what I judge, and it is my own law because I judge it.

For radical Kantian constructivists, neither response will be sufficient. Moderate constructivism is an incomplete account of practical principles that presupposes an unvindicated kind of practical reasoning. Whereas radical Humean constructivism is committed to an individualistic relativism that makes practical reason subservient to unsupported judgements about reasons and decouples it from moral universalism. Instead, Korsgaard and O'Neill's radical Kantian constructivisms hope to avoid the spectre of arbitrariness through appeal to internal norms of self-legislation which are presupposed by the perspective of any potential legislator. Thus, although Korsgaard recognises the threat of ‘arbitrary power’ to undermine self-legislation, and O'Neill warns of ‘anomie’ and ‘disorientated consciousness’, they are confident that the deep structure of practical rationality can be mined for resources to guide self-legislation and stave off arbitrariness.21 Our task will be to determine whether they succeed.

V

Korsgaard abandons Rawlsian neutrality about fundamental sources of normativity, claiming that ‘values are created by human beings’ and this is ‘a matter of making laws’.22 Her ambitious radical constructivist arguments have been much-discussed, and among many targets of criticism are her transcendental defence of humanity's value, constitutivist conception of agency and motivation, and fusion of ethical and metaethical reasoning.23 I do not want to linger on this well-ploughed ground, so shall focus only on the most direct threats from the paradox of self-legislation which confront her constructivism.

Korsgaard's answer to why we should legislate for ourselves is that it solves a problem that humans cannot escape. This problem arises from the nature of agency, and we humans ‘are condemned to choice and action’.24 The ever-present capacity for reflection—to ask whether there is a reason to do as we are inclined—is meant to reveal that the self is divided. We are sundered by self-consciousness, since awareness of potential motivations creates an inner distance from them that precludes mere instinctive behaviour. Deliberation attempts to reunite the self under a common purpose; and a deliberative choice require a principle. Yet, the constitution of agency cannot begin from merely any principle—it requires ‘not one imposed on it from outside, for it has no reason to accept such a principle, but one that is its own’.25 Thus, self-legislation is demanded by agency itself.

This argument for self-legislation begins from a psycho-philosophical characterisation of agency. Action presupposes maxim-formation, through which the will freely adopts action-guiding principles which provide it with reasons.26 Two claims concerning self-determination are introduced here: first, that the rational will cannot be determined by incentives which are not incorporated into maxims it adopts; and second, that the rational will can be determined by incentives which it does incorporate into maxims it adopts. In short, insofar as action is necessary, the first negative claim implies the second positive claim. Thus, if action requires principles, and you have to act, all whilst being unable to act on principles that you do not give to yourself, then it follows you must act on principles that you do give to yourself.27

What makes this a Kantian rather than Humean radical constructivism is the further argument that legislators are committed to some common values, such as the value of human rational agency itself. The argument is premised on the claim that not all the contingent practical identities which we use to orient our deliberation—brother, vegetarian, philosopher, and so on—can be shed without leaving us bereft of reasons and so unable to act. In recognising this fact—that we must value ourselves under contingent conceptions of our identity—we are valuing something non-contingent: ourselves as rational agents. This is because ‘we are endorsing a reason that arises from our rational nature—namely our need to have reasons’.28 This means that self-legislation cannot be entirely arbitrary, as if subject only to inter-attitudinal constraints, because all legislators must be committed to valuing their rational agency.

VI

The main objection to Korsgaard which I shall press concerns her denial that the rational will can be legitimately and intelligibly subject to non-self-legislated normative authorities. Korsgaard directs this claim against her non-constructivist rivals. In doing so, she characterises scepticism as ‘the fear that we cannot find what Kant called “the unconditioned” ’, namely an answer to reflective scrutiny that ‘makes it impossible, unnecessary, or incoherent to ask why again’.29 Korsgaard wields this challenge against rival non-sceptical accounts of normativity, claiming that ‘[t]he realist's response is to dig in his heels’, ending any threatened justificatory regress by fiat, through an appeal to intrinsically normative entities.30 However, her need to hold non-constructivist competition to such high standards redounds against her.

Kant, of course, takes the need for relentless criticism seriously, proclaiming that, ‘Our age is, in especial degree, the age of criticism, and to criticism everything must submit’.31 However, Korsgaard herself is ill-equipped for this critical task. She only gives us reasons to treat some of our practical identities as reason-giving, and this falls short as guidance to legislators given her own strictures on reflection, on which practical reasoning requires an ‘unconditional’ answer to legitimately conclude it. For it is one thing to say that we should treat some contingent identities as reason-giving, but another to say which we ought to. Demonstrating why we must legislate is of limited use in determining how to legislate.

On Korsgaard's own demanding view, we are not to cease reflection until it is ‘impossible, unnecessary, or incoherent’ for us to continue. But it seems possible, necessary, and coherent to reflect about which specific identities to adopt, as distinct from reflecting about the necessity that, as a condition of agency, we must adopt one or other identity.32 The paradox of self-legislation will not be solved until we can find normative guidance for reflection at this more fine-grained level. An agent needs some way of determining which identities could be adopted consistently with non-arbitrary self-government: the bare fact of our reflective endorsement of them is not enough if there are insufficient rational constraints upon what to endorse. Korsgaard offers us no help with this task, and her own requirements of unconditionality in reflection make it a particularly onerous project. In this respect, she fails to dissolve the paradox.

In response, Korsgaard might urge us to take a more variegated approach to the vindication of practical identities. The moral law enjoins us to value humanity, and we can distinguish three types of practical identity in relation to this moral law: those which are permissible, impermissible, and mandatory. In line with these distinctions, Korsgaard could say that it is only in assessing whether an identity is morally mandatory or morally impermissible that we are required to ask whether it would be incoherent either to adopt or not adopt it. In between these extremes, there will be a whole raft of acceptable but discretionary identities—e.g. student of German, casual acquaintance, hip-hop fan—which would give us reasons in light of our endorsement of them and their interaction with our other self-conceptions. To some extent these middling identities would partake of our moral identities, because these moral identities include the recognition that we should adopt a set of contingent self-conceptions in navigating our lives. But it would still be coherent to reject many of these identities, since they are not imposed directly by the moral law but only indirectly by the need to endorse some reason-conferring outlook. The hope would be to avoid an overly demanding set of standards for our ordinary non-moral identities by subjecting them only to the test of moral permissibility and consistency with our other non-moral identities. At the same time, genuinely universal rational constraints stemming from our autonomy as agents would be in force (in the form of moral norms), providing sufficient structure and guidance to enable people to gain enough critical purchase on their lives to live them independently and avoid completely capricious self-legislation.

The problem with responses of this kind is that they introduce an unstable mix of rigorist and voluntarist elements to practical reasoning, such that the content of the will is sovereign once the demands of morality are satisfied. The commonest sort of critical scrutiny which we find ourselves engaging in as practical agents does not concern morality directly. Rather, in Korsgaard's idiom, it will be negotiating our non-moral practical identities in determining how we should act. This task requires people to weigh their contingently incurred commitments, judging which are most pressing, what trade-offs must be made, how they should go about formulating a plan of action, and so on. But when it comes to these morally permitted but not morally mandatory matters, then the excoriating critical force of the moral law cannot guide us.

In the absence of the moral law, then the rational constraints on deliberation are excessively weak. Other than mere consistency, there are slim resources for distinguishing non-moral identities which are valuable guides from those which can be readily discarded. Similarly, even within the set of identities we have adopted, the same difficulty arises for determining what takes precedence as a source of reasons in case of conflict or underdetermination. Of course, nothing much hangs on many intentional actions, and some are perhaps not even amenable to ratiocination or reason. It will also very often be the case that endorsement of an action in line with our consistent desires and commitments is all the reason we could need (e.g. deciding to act on a desire to read W.G. Sebald during your train journey rather than Iris Murdoch is likely sufficient reason to do so). Yet, moral necessities and consistent preference-endorsement do not exhaust the space of practical reason. To take one kind of example, when evaluating our character and nexus of desires as a whole—asking ourselves if we are the sort of person we ought to be—then independently compelling political, social, aesthetic, and intellectual considerations can legitimately sit alongside those of morality (even if the latter are categorically binding).

When reasons give out too soon, we are left with only a quasi-rational legislation of the will. If, within the bounds of the moral law, only our endorsement of non-moral identities matters, then too large a part of our life is surrendered to those conditions under which our non-moral preferences are formed.33 This route would lead Korsgaard back towards the difficulties we met with Humean forms of radical constructivism: practical reason is held hostage to what we simply happen to care about rather than being able to help us work out what that should be. There should be room for discretionary activity and evaluation in the autonomous life, but when an account of self-legislation threatens to deprive us of the resources to critically appraise much of the rest of our activities and evaluations, then it fails to do justice to our autonomy as a whole.

VII

The radical Kantian constructivism which O'Neill advances is more modest in its aims than Korsgaard's but promises to be more stable and better-equipped to defuse the paradox of self-legislation. O'Neill gives a ‘constructive account of practical reason and ethics’, yet without claiming that all values are constructed.34 Despite this, she is still a radical constructivist, because there are no vertical limits to construction—it is not supported or oriented by non-constructed laws or values.

O'Neill has a distinctive understanding of self-legislation: rather than individual agents, reason itself figures as the legislator. Reason is self-legislative because it derives its authority reflexively rather than from an external source.35 Nevertheless, self-legislation remains a feature of human life, as ‘the basic characteristic of ways of thinking or willing that are conducted with sufficient discipline to be followable by or accessible to others’.36 Yet, unlike Korsgaard, O'Neill is not committed to the claim that our reflective endorsement is the source of normativity. This results in a more pronounced rationalism that chimes with critics of Korsgaard such as G.A. Cohen, who claims that ‘it is reason as such that is sovereign over us, and that gives determinacy, stability, and authority to a law that would otherwise lack all that’.37

For O'Neill, reason is identified with the autonomy of thought and action.38 Its authority consists in a form of self-discipline, following Kant's claim that

Reason must subject itself to criticism; should it limit freedom of criticism by any prohibitions, it must harm itself, drawing upon itself a damaging suspicion.39

The principles of rationality are simply those accorded provisional authority in virtue of their survival of self-criticism and dependence upon no alien authority.40

O'Neill's justification for self-legislation centres on the need to avoid deference and anomie in reasoning.41 Reasoning that defers to other authorities remains merely conditional, and without vindicating its antecedents it ultimately destroys itself, since reason's claim to legitimacy can only ever be its own autonomy—the ability to vindicate itself. Whereas anomic reasoning, which defers to no other authorities but which gives itself no law either, leads to ‘incoherence and isolation’, since this reasoning is paralyzed by a lack of structure and cannot be communicated to others.42 Therefore, without self-legislation, reasoned thought or action cannot arise. This answers the first problem raised by the paradox of self-legislation, namely why we should give any sort of law to ourselves.

We must next ask how we should do so. As we have already seen, reasoners are meant to be unable to rely upon authorities unvindicated by reason.43 O'Neill gives this description of reason's situation:

One corollary of refusal to bow under an alien yoke is that what count as principles of reason cannot hinge on variable and contingent matters, all of which, however intimately human, are alien causes.44

The negative demand to eschew alien causes can be transformed into positive advice: the norms we adopt must hold irrespective of our contingent circumstances. This makes the fundamental principle of reasoned thought and action into a form of the categorical imperative, which directs our will according to principles that could be laws for all. Consequently, ‘ “self-legislation” is not a mysterious phrase for describing merely arbitrary ways in which a free individual might or might not act’; it is conformity to laws which must be ‘in principle intelligible to others, and open to their criticism, rebuttal or reasoned agreement’.45 Further arguments are adduced in an attempt to show that we can get concrete practical guidance from these conditions, which prohibit activities like direct and indirect injury, deception, and an attitude of ethical indifference.46

VIII

Despite its attractions, O'Neill's radical constructivism tries to do too much with too few resources. The tools which she allows herself are an ‘empirically realistic view of the capacities and capabilities agents have,’ alongside abstraction from this starting point.47 This precludes building idealised models of humanity and the human situation that are shaped by specific values. Rawls' Kantian ideal of the person as autonomous would count as one such idealisation which O'Neill is unhappy with and which she rejects because it forsakes a neutral description of agents, such that it is in fact false of some of those to whom it is meant to apply.48 The only additional element of O'Neill's constructivism is the constructive procedure. Again, this is characterised modestly, such that it amounts to no more than the requirement to employ ‘procedures thought to be followable by others’.49 Any account of reason which could successfully build upon such minimal and difficult to contest foundations would be extremely powerful. However, I think O'Neill fails in this task.50

On O'Neill's radical Kantian constructivist model of self-legislation, it is a matter of combining suitable materials of construction with a constructive procedure. For her, the materials of construction are simply empirical facts considered under various levels of abstraction, and so in legislating action-guiding principles, the function of the constructive procedure must be to introduce a normative dimension to these empirical facts. This raises the question of what justifies the constructive procedure itself. O'Neill takes the procedure of construction to be vindicated recursively. In other words, it is justified through its ability to meet the same standards—namely, being able to be followed by others in thought and action—which it prescribes as a norm for deriving other norms. This is combined with the claim that other accounts of practical reasoning cannot meet these minimal rational standards. Her strategy relies upon a via negativa, which seeks to rule out competing conceptions of practical reason by demonstrating their inability to meet minimal standards of rationality which her Kantian constructivism can.

O'Neill's response to rival models of practical reason—teleological, instrumental and particularist—is instructive. All three are accused of being at worst incomprehensible and at best conditionally reasoned (and therefore arbitrary). For instance, teleological models of practical reason, which suppose we can discover what is good, are ruled out because they would require contentious metaphysical support to show that there was a non-constructed good. The problem is located not in any substantive argument with the advocate of teleological reason—disagreement over the plausibility of this or that metaphysical commitment—but the formal property of needing metaphysical foundations at all. The supposed failure of teleological accounts to meet the condition of being followable by all in thought and action is because ‘those who do not accept the appropriate metaphysical and other arguments and positions [enabling knowledge of the objective good] may find proposals that depend upon them at worst incomprehensible and at best conditionally reasoned’.51 The implicit parallel drawn is between metaphysical argumentation and private uses of reason in Kant's sense.52

Given O'Neill's modal formulation of the conditions of followability, it is strange that here she relies on the fact that people might, but again might not, find teleological accounts unintelligible. Why is it not enough that people could understand such an account for it to allow teleological reason to be followable in thought and action? Unless O'Neill shows that a requirement for such accounts to draw upon metaphysical support makes them unintelligible in principle for beings like us, which would be a demanding task, then disqualifying them on these grounds appears too hasty. However, even if objections on the grounds of potential unintelligibility are too strong, then O'Neill could still claim that teleological accounts are conditionally reasoned, requiring the vindication of assumptions about values or wider reality which, even if they do happen to be true, require support from outside of reason for us to justify. On her stringent conception of self-legislation, this would make adherents of teleological reasoning heteronomous, insofar as they attempt to act upon conditions alien to the will which give it an object, rather than allowing it to legislate for itself on the grounds of reason.

If O'Neill's strategy for dismissing teleological accounts of reasoning turns on the objection that such reasoning will be conditioned, then whatever the success of her objections to rival accounts, this introduces vulnerabilities into her own account. Can it escape the charge of being equally conditioned? That is, does it ground reasoning upon contingent assumptions, and so make what we have reason to do dependent upon some further condition which may or may not hold? I think that the demanding standards which O'Neill must introduce to dismiss rival accounts of practical reason leave her susceptible to this objection. Her account of reasoning and the justification of the constructive procedure is particularly vulnerable, since it relies upon co-ordination and communication amongst agents figuring as goals. Indeed, O'Neil's fundamental principle of rationality—‘to reject principles and strategies that are not followable by all'—derives its authority as a response to an anterior ‘coordination problem’ facing agents.53 But once her critique of rival accounts of reasoning has introduced a prohibition on conditioned forms of reasoning, then presupposing this problem and these goals begins to become problematic.

If co-ordination and communication are not the ends of all people, then it seems that O'Neill's account of reason is merely hypothetical. Given that you want to live a life in dialogue with a plurality of people with whom you can engage in a series of reciprocal interactions, then you ought to adopt the principle that the maxims you adopt should be followable in thought and action by all. This is a perfectly reasonable stretch of ordinary practical reasoning, but will not do for a vindication of the procedure of construction (at least on the basis that O'Neill wants). This is because it makes reason beholden to what she has called ‘private’ authorities: those which are dependent upon us already accepting the assumptions of some contingent set of plans, institutions, or forms of life.

Co-ordination and communication are goals amenable to abstract characterisation that can demonstrate that the vast majority of people implicitly or explicitly accept them in some form. But the radical aspirations of O'Neill's constructivism have seen her reject a restricted construction of normativity which relativise the authority of reason to what a subset of people happen to accept. All agents—or all competent and capable ones—were supposed to fall within the bounds of reason. Yet, for all in her thought that warns against it, O'Neill is in danger of instrumentalising reason, insofar as it is put into the service of a positive desire to coherently interact and communicate with other people. No doubt a human life without the implicit goal of such interaction and communication would be paltry; but its possibility seems to demonstrate that, for O'Neill, reasoning has a minimally hypothetical form. Therefore, by her own lights, she ought to fall short of vindicating self-legislation.

IX

In reply, co-ordination and communication might be said not to be empirically discernable aims of reasoners but integral goals of reasoning as such. O'Neill has forsworn idealisation of agents—which attributes properties to them which some do not possess—so non-empirical goals of reasoning could not be understood as ideal goals of agents. Instead, these goals of co-ordination and communication might be commitments which legislators incur in the totality of their willing, akin to those which she labels ‘Principles of Rational Intending’.54 O'Neill claims that human beings invariably have at least some projects or maxims which cannot always be realised unaided; and this suggests that legislators always incur commitments to goals of co-ordination and communication with others, whether or not they actually recognise them as commitments.55 Therefore, these goals for orienting legislation would be neither empirical nor ideal but practically necessary.

This response depends upon vindicating the claim that all rational agents have projects which ensure that they are dependent enough on others to impose a commitment to co-ordinate and communicate with them. Can reasoners never adopt a self-sufficient set of maxims? This question concerns the ends of reason as such rather than the contingent projects of actual agents. The more ambitious claim must be defended because rational self-legislation whose stability is dependent upon specific projects or commitments being adopted which reasonable agents could legitimately reject would rely on conditioned reasoning. Therefore, self-legislation must rest on claims about what all reasoners must legislate, whatever other commitments they happen to incur. Yet, even the impressive principles of rational intending that O'Neill amasses give no clear way to argue that the goals of co-ordination and communication which could orient self-legislation are sufficiently universal that they are integral to reasoning as such. Consequently, the ambitious thesis about the ends of reason remains undefended.56

O'Neill's recursive strategy for vindicating constructivist models of self-legislative practical reasoning does not meet the challenge that she sets for rival accounts of practical reason. She transgresses her own necessarily strict requirement that self-legislation be independent of conditioned forms of reasoning, where there are no grounds for justifying genuinely practically necessary standards of reason to guide legislation. Thus, O'Neill does not escape the paradox of self-legislation.

X

The radical Kantian constructivist accounts we have considered have failed to offer a workable conception of self-legislation. Neither rational agents nor the structure of rational agency itself can be the ultimate source of legislation, since neither approach adequately accounts for the normative authority needed to prevent legislation collapsing into caprice or other forms of arbitrariness. The main alternative approach to self-legislation within metaethics is radical Humean constructivism, which understands legislation relativistically. Yet, relativism about value is a high price to pay to ground normativity in human autonomy.57 Therefore, self-legislation may seem like a moribund metaethical concept. However, there is a neglected self-legislative tradition that predates Kant and his constructivist readers. Unearthing and building upon this early conception of self-legislation shows how the metaphor can be rehabilitated and put to work outside of constructivist metaethical projects.

Thought about self-legislation can be traced at least as far back as ancient Greek discussions of αύτονομία (autonomy). On a literal reading, to be autonomous—‘αύτο’ meaning self or one's own and ‘νόμος’ meaning law or custom—is simply to be self-legislating, in the sense of having a law of one's own. The main use of ‘αύτονομία’ was in Greek political vocabulary, denoting minor states retaining relative independence from their more powerful allies, and who remained subject to their own laws rather than those of a foreign power.58 However, its first extant appearance is in Sophocles, who ascribes autonomy to an individual. The Chorus describe Antigone as autonomous in her descent into Hades which follows the discovery of her attempted burial of her brother. It suggests that her avowed piety in fulfilling the divine law is an act of fidelity to a law of her own.59

Subsequently, autonomy is praised by some Greek philosophers writing under the influence of Stoicism, such as Epictetus and Dio of Prusa.60 Autonomy is once more understood as adhering to a law of one's own which is also divine, eternal and natural. The wise achieve freedom by knowing and rigidly holding to this natural law, which allows them to escape the kind of determination by contingent circumstances that chasing pleasure or status involves.61 The operative sense of self-legislation in these attributions of individual autonomy is appropriation of an already-authoritative law, whose content the individual does not choose or otherwise directly determine.

We find a Christianised version of this kind of self-legislation—as adherence to natural law—in Aquinas' commentary on St. Paul's Epistle to the Romans. Paul says that conscience and reason allow Gentiles to be ‘a law unto themselves’.62 On Aquinas' reading, this means that rational creatures are a law to themselves insofar as they recognise the natural law.63 Guiding oneself toward this law by means of reason is said to be ‘the highest degree of worth in human beings, namely that they are led not by others but by themselves towards good’.64

Christian Wolff's perfectionist articulation of the natural law tradition develops this conception of self-legislation.65 Wolff claims that reason enables us to know a law of nature obligating us to make ourselves and others more perfect. Echoing Paul and Aquinas, he says:

Because we know through reason what the law of nature requires, a reasonable man needs no further law, for because of his reason he is a law unto himself.66

On this model of self-legislation, rational cognisance of a natural law can transform it into a law of one's own. The rationalist foundations of self-legislation are significant because reason is a distinctive influence upon human action that can guide people without doing so ‘through compulsion, like a beast’.67 Furthermore, the reasonable person neither has nor needs a ‘superior’ to command them (and so ‘becomes like God’).68 In contrast, the unreasonable person is only motivated to do what is good out of fear of punishment or hope of reward, and so remains in childlike submission to others. The law that the reasonable person appropriates is normatively realist in character: it is ‘natural’ in the sense of being ‘validated by nature itself’ in contradistinction to human positive law or divine command.69 Thus, Wolff provides a schema for understanding self-legislation which is realist and recognises both reason's non-coerciveness and its ability to liberate people from subjection to the authority of others.70

XI

I have argued that radical Kantian constructivism runs aground by taking self-legislation to be the ultimate source of normative authority in practical reasoning. This diagnosis suggests that a viable conception of self-legislation must abandon a radical constructivist metaethical underpinning. We have seen that there is already a historical precedent for realist accounts. In positing normative authorities that obtain anterior to self-legislation, then the realist has the resources to avoid the paradox which radical constructivist metaethics introduces. In concluding, I shall briefly outline how a rationalist and realist approach to self-legislation might nonetheless perform analogous functions to its radical constructivist rivals.

Consider the pitfalls of anomie and deference again, which O'Neill intends self-legislation to overcome. The anomic agent has no authoritative structures to govern their actions; whereas, the deferential agent is governed by alien authorities. Yet, rational appropriation of independently authoritative laws, such as ethical principles, would provide no less structure to agency than self-authored laws. Therefore, listless or wanton anomie is avoided. Forestalling deference presents more difficulties. Realist self-legislation does not need to directly invoke another human or divine will in explaining normative authority or the motivation to follow laws. However, the deeper threat still concerns heteronomy: the objection that, as practical reasoners, we would not give the law to ourselves but find it outside of us.

How could obeying an exogenous law amount to self-determination? In reply, we should disaggregate whether a law is alien from the source of its normative authority. Kant himself recognises that legislators need not be originators or authors of the law.71 To legislate is to command that a law be followed; and this does not preclude the law arising or drawing its justification from elsewhere. In short, legislation in which laws become one's own is an achievement, which is secured through sensitivity to the reasons why these laws are authoritative. This rational self-legislation is a reckoning with normative authority rather than an institution of it.

If one of the main functions that an account of self-legislation performs is making sense of threats to agency which kinds of alienation impose, then this can indicate some limits to rational self-legislation. Whether or not we rationally legislate is only one relevant factor in determining whether a law is sufficiently alien to make us heteronomous. On a narrow model of rationality, legislation through cognitive acknowledgement of the reasons justifying a law needs to be supplemented by a process of affective and volitional harmonisation with it. This is because remaining deeply resistant to a law which we think we should act on is a form of estrangement from ourselves.72 To prevent them remaining alien, legislated norms would have to be integrated into our character and emotional life. This process can reach a point where the legislative metaphor appear less apt because a norm is no longer experienced as traditionally law-like (e.g. as a stern imperative which categorically obliges us). Similarly, an implicitly deontic framework might always be ill-fitting when discussing some subsets of practical norms, such as etiquette or virtues like generosity. Nevertheless, for a significant range of normative phenomena, realist understandings of rational self-legislation may be able to provide a promising alternative to radical constructivist strategies for avoiding anomie and deference in practical agency.

In gesturing towards a realist account, I have built upon a diagnosis of constructivist attempts to exploit self-legislation. We saw how contemporary conceptions of self-legislation have their roots in a radicalisation of the Rawlsian project of political constructivism. In superseding Rawls, while avoiding the relativism of their Humean cousins, radical Kantian constructivists had to show how self-legislation could be the fundamental source of practical normativity without collapsing into arbitrary legislation. I argued that both leading attempts to do so founder in attempting to meet the strict success conditions their approaches impose. Korsgaard cannot adequately justify the critical resources which agents use to navigate their practical identities. This leaves her account riven between rigorism and voluntarism, such that it will not escape the paradox that arises when self-legislation is unable to appeal to external normative standards. O'Neill anchors self-legislation more firmly to the self-disciplining structures of reason itself. However, she ultimately fails to defend sufficiently unconditional practical norms which could guide legislation. The endemic problems with radical constructivist models of self-legislation prompted a reconstruction of a neglected realist tradition of which Wolff's position presents a paradigm example. In outlining a rationalist and realist account of self-legislation, we saw that it could also make sense of our ability to overcome anomie and deference in practical action. Thus, we need not make laws but can make them our own.73

Notes

  1. 1

    The locus classicus is found in the Groundwork to the Metaphysics of Morals: ‘the will is not merely subject to the law but subject to it in such a way that it must be viewed as giving the law to itself and just because of this as first subject to the law (of which it can regard itself as the author)’. Immanuel Kant, Practical Philosophy ed. and trans. Mary Gregor (Cambridge: Cambridge University Press, 1999), p. 81, Ak 4:431.

  2. 2

    There is no shortage of critics who challenge the Kantian credentials of Rawls and his constructivist successors. See Otfried Höffe, ‘Is Rawls' Theory of Justice Really Kantian?Ratio 26 (1984), pp. 103124; Larry Krasnoff, ‘How Kantian is Constructivism?Kant-Studien 90:4 (1999), pp. 385409; Patrick Kain, ‘Realism and Anti-Realism in Kant's Second CritiquePhilosophy Compass 1:5 (2006), pp. 449465; Kerstin Budde, ‘Rawls on Kant—Is Rawls a Kantian or Kant a Rawlsian?European Journal of Political Theory 6:3 (2007), pp. 339358. On the interpretation of self-legislation in particular, see Patrick Kain, ‘Self-Legislation in Kant's Moral PhilosophyArchiv für Geschichte der Philosophie 86 (2004), pp. 257306. I think these non-constructivist readings of Kant are more accurate, but I will bracket these exegetical disputes in what follows.

  3. 3

    John Rawls, A Theory of Justice, Revised Edition (Cambridge, MA: Harvard University Press, 2001), p. 225.

  4. 4

    John Rawls, ‘Kantian Constructivism in Moral Theory’, Journal of Philosophy 77: 9 (1980), p. 543 and Political Liberalism (New York: Columbia University Press, 1993), p. 32.

  5. 5

    For arguments against construing constructivism in proceduralist terms, see Sharon Street, ‘What is Constructivism in Ethics and Metaethics?Philosophy Compass 5:5 (2010), pp. 363384.

  6. 6

    John Rawls, Political Liberalism, p. 96; cf. Christine Korsgaard, The Sources of Normativity, with commentary by G.A. Cohen et al. and an introduction by Onora O'Neill (ed.) (Cambridge: Cambridge University Press, 1996), pp. 3637.

  7. 7

    John Rawls, ‘Themes in Kant's Moral Philosophy,’ (1989) reprinted in Collected Papers, ed. Samuel Freeman (Cambridge, MA: Harvard University Press, 1999), p. 514.

  8. 8

    Rawls' later clarification of his position is well-summarised by O'Neill: ‘Far from deriving a justification of democratic citizenship from metaphysical foundations, Rawls invites us to read A Theory of Justice as a recursive vindication of those deep principles of justice “we” would discover in drawing on “our” underlying conceptions of free and equal citizenship. This vindication of justice does not address others who, unlike “us”, do not start with such ideals of citizenship. It has nothing to say to those others. It is “our” ideal, and “our” justice’. Onora O'Neill, Constructions of Reason, p. 211. On the development of Rawls' constructivism, see Onora O'NeillConstructivism in Rawls and Kant,’ in The Cambridge Companion to Rawls, ed. S. Freeman, (Cambridge: Cambridge University Press, 2002). For criticism of Rawls' civic turn and a defence of the universalist project implicit in his earlier and supposedly more Kantian work, see Robert Taylor, Reconstructing Rawls: The Kantian Foundations of Justice as Fairness (University Park, PA: Pennsylvania State University Press, 2011).

  9. 9

    See John Rawls, Political Liberalism, p. 95.

  10. 10

    Other moderate forms of constructivism which forsake Rawls' metaphysical neutrality about reason (being metanormatively realist and metaethically expressivist, respectively) are advanced in Tim Scanlon, What We Owe to Each Other (Cambridge, MA: Harvard University Press, 1998) and James Lenman, ‘Humean Constructivism in Moral TheoryOxford Studies in Metaethics: Volume 5 (Oxford: Oxford University Press, 2010), pp. 175193.

  11. 11

    For example, on the relation between value, constructivism and self-legislation, Korsgaard writes: ‘Values are not discovered by intuition to be “out there” in the world […] values are constructed by a procedure, the procedure of making laws for ourselves’. Korsgaard, The Sources of Normativity, p. 112. In contrast, Rawls claims: ‘we do not say that the procedure of construction makes, or produces, the order of moral values’. Rawls, Political Liberalism, p. 95. His political liberalism rejects this kind of ‘constitutive autonomy’, which he associates with Kant's ‘transcendental idealism’ (ibid. pp. 99100).

  12. 12

    For discussion of this paradox in a historical context, see Robert Pippin, ‘The Actualization of Freedom’, in K. Ameriks (ed.), The Cambridge Companion to German Idealism (Cambridge: Cambridge University Press, 2000), pp. 180199; Terry Pinkard, German Philosophy 1760–1860: The Legacy of Idealism (Cambridge: Cambridge University Press, 2002); Robert Pippin, Hegel's Practical Philosophy: Rational Agency as Ethical Life (Cambridge: Cambridge University Press, 2008). Critical discussion can be found in Robert Stern, ‘Pinkard on German Idealism’, Bulletin of the Hegel Society of Great Britain 49 (2004), pp. 1–17 and ‘Freedom, Self-Legislation and Morality in Kant and Hegel: Constructivist vs Realist Accounts’, in E. Hammer (ed.) German Idealism: Contemporary Perspectives (London: Routledge, 2007), pp. 245266.

  13. 13

    Donald Regan, ‘The Value of Rational Nature’, Ethics 112 (2002), p. 278.

  14. 14

    For an intramural discussion of whether this ‘why?’ question should be answered by constructivists, see Sharon Street, ‘Coming to Terms With Contingency: Humean Constructivism About Practical Reason’, Constructivism in Practical Philosophy, eds. James Lenman and Yonatan Shemmer (Oxford: Oxford University Press, 2012), pp. 4059.

  15. 15

    Korsgaard explicitly rejects a similar appeal to a ‘heroic existentialist act’ (albeit one that concerns the contingent content of the will rather the choice of principles binding it). See Christine Korsgaard, ‘The Normativity of Instrumental Reason,’ Ethics and Practical Reason, eds. Garrett Cullity and Berys Gaut (Oxford: Clarendon Press, 1997), p. 252.

  16. 16

    On the contrast with radical constructivism's commitment to constructing all normative authority, see Onora O'Neill, ‘Constructivism vs. Contractualism’, Ratio 16:4 (2003), pp. 319331 and Christine Korsgaard, ‘Realism and Constructivism in Twentieth-Century Moral Philosophy’, pp. 117118.

  17. 17

    John Rawls, Political Liberalism, p. 26.

  18. 18

    If, alternatively, self-legislation is construed as an additional yet not fundamental source of normativity for the moderate constructivist, and thus derivatively contributes to ‘creating’ certain norms, then the dilemma looks to re-emerge, albeit in a weakened form. To the extent that prior norms contribute to determining what is to be legislated, then legislation is redundant, since it is those norms that are authoritative; to the extent that other factors do so, then self-legislation is not fully rational, in ways that reintroduce concerns about its ungrounded and thereby arbitrary nature. Thus, any decrease in redundancy would carry a cost in arbitrariness and vice versa. However, I shall not rely upon or pursue this objection to these forms of moderate constructivism any further.

  19. 19

    The term is introduced in Carla Bagnoli, ‘Moral Constructivism: A Phenomenological Argument’, Topoi 21 (2002), p. 131. See also Valerie Tiberius, The Reflective Life: Living Wisely With Our Limits (Oxford: Oxford University Press, 2008), ch. 8.

  20. 20

    Sharon Street, ‘Constructivism about Reasons’, p. 224.

  21. 21

    Korsgaard, ‘Realism and Constructivism in Twentieth-Century Moral Philosophy’, Philosophy in America at the Turn of the Century: APA Centennial Supplement to The Journal of Philosophical Research (Charlottesville, Virginia: The Philosophy Documentation Center, 2003), p. 114; Onora O'Neill, Autonomy and Trust in Bioethics (Cambridge: Cambridge University Press, 2002), p. 92 and Constructions of Reason, p. 3.

  22. 22

    Korsgaard, The Sources of Normativity, p. 112.

  23. 23

    For representative criticisms of Korsgaard on these points, see Robert Stern, ‘The Value of Humanity: Reflections on Korsgaard's Transcendental Argument’, Transcendental Philosophy and Naturalism eds. Joel Smith and Peter Sullivan (Oxford: Oxford University Press, 2011), pp. 7495; David Enoch, ‘Agency, Shmagency: Why Normativity Won't Come from What Is Constitutive of Agency,’ Philosophical Review 115: 2 (2006), pp. 169198; and Nadeem Hussain and Nishi ShahMisunderstanding Metaethics: Korsgaard's Rejection of Realism’, Oxford Studies in Metaethics: Volume 1 ed. Russ Shafer-Landau (Oxford: Oxford University Press, 2006), pp. 265294.

  24. 24

    Christine Korsgaard, Self-Constitution (Oxford: Oxford University Press, 2009), p. 1. She goes on to say (p. 2) that ‘The necessity of choosing and acting is not causal, logical, or rational necessity. It is our plight: the simple inexorable fact of the human condition’.

  25. 25

    Korsgaard, Self-Constitution, p. 213.

  26. 26

    Korsgaard, The Sources of Normativity, p. 94.

  27. 27

    On the ambiguous status of this must, see Enoch, ‘Agency, Shmagency’, pp. 188189, n. 42.

  28. 28

    Korsgaard, Self-Constitution, p. 24.

  29. 29

    Korsgaard, The Sources of Normativity, p. 94 and p. 33.

  30. 30

    Korsgaard, The Sources of Normativity, p. 30.

  31. 31

    Kant, Critique of Pure Reason, Axi.

  32. 32

    Korsgaard seems to concede this when she acknowledges that her account of practical reason suffers from a ‘missing principle’. See Self-Constitution, pp. 5253.

  33. 33

    Korsgaard sometimes even suggests that our actual endorsement is required to secure the authority of the moral law itself. See Korsgaard, The Sources of Normativity, p. 254. Cf. Christopher Gowans, ‘Practical Identities and Autonomy: Korsgaard's Reformation of Kant's Moral Philosophy,’ Philosophy and Phenomenological Research 64: 3 (2002), pp. 546570; p. 556.

  34. 34

    Onora O'Neill, Towards Justice and Virtue: A Constructive Account of Practical Reason (Cambridge: Cambridge University Press, 1996), p. 211.

  35. 35

    ‘A better reading of Kant's idea of self-legislation would view the element self in self-legislation simply as a reflexive term.’ Onora O'Neill, Autonomy and Trust in Bioethics, p. 85.

  36. 36

    O'Neill, Autonomy and Trust in Bioethics, p. 89.

  37. 37

    Cohen in Korsgaard, The Sources of Normativity, p. 172, emphasis mine.

  38. 38

    O'Neill, Constructions of Reason, p. 64.

  39. 39

    Kant, Critique of Pure Reason, A738/B766, quoted in O'Neill, Constructions of Reason, p. 57.

  40. 40

    O'Neill, Constructions of Reason, p. 58.

  41. 41

    Onora O'Neill, Autonomy and Trust in Bioethics (Cambridge: Cambridge University Press, 2002), p. 93. Note that these parallel the threats of self-defeating redundancy and arbitrariness that the paradox of self-legislation poses.

  42. 42

    O'Neill, Autonomy and Trust in Bioethics, p. 94.

  43. 43

    O'Neill, Constructions of Reason, p. 64.

  44. 44

    O'Neill, Constructions of Reason, p. 58.

  45. 45

    O'Neill, Autonomy and Trust in Bioethics, p. 95.

  46. 46

    O'Neill, Towards Justice and Virtue, ch. 6.

  47. 47

    O'Neill, Bounds of Justice, p. 7.

  48. 48

    See the criticism of Rawls in her ‘Constructivism in Ethics’, Constructions of Reason, p. 206f.

  49. 49

    O'Neill, Towards Justice and Virtue, p. 210.

  50. 50

    This analysis is significantly indebted to Kerstin Budde's perceptive discussion of O'Neill in ‘Constructivism All The Way Down—Can O'Neill Succeed Where Rawls Failed?Contemporary Political Theory 8 (2009), pp. 199223.

  51. 51

    O'Neill, Towards Justice and Virtue, p. 51.

  52. 52

    See Immanuel Kant, ‘An Answer to the Question: What Is Enlightenment?Practical Philosophy trans. and ed. Mary Gregor (Cambridge: Cambridge University Press, 1996), pp. 1722.

  53. 53

    O'Neill, Towards Justice and Virtue, p. 60.

  54. 54

    For instance, the hypothetical imperative, which requires willing the means to our ends, would count as such a principle of rational intending. O'Neill, Constructions of Reason, p. 91; cf. p. 96.

  55. 55

    O'Neill, Constructions of Reason, p. 101.

  56. 56

    Another way of interpreting O'Neill's arguments for constructivism would be to deny that the problem of co-ordination has the foundational role that critics such as myself and Budde identify and instead read her strategy as closer to Korsgaard's reconstruction of the derivation of the categorical imperative in the Groundwork. Such an argument would begin with a conception of agency, determine that a principle was needed to structure the will, and conclude that the only principle that all agents could follow is one that requires acting only on principles followable by everyone. However, O'Neill would have problems relying solely on such an argument, because a conception of agency that would be robust enough to allow such a derivation would seem to presuppose the very idealisations of our situation as agents that O'Neill wants to depart from. Yet, this rejection of idealisation is among her most distinctive contributions to constructivism, setting her apart from figures such as Korsgaard and Rawls, as noted by commentators such as Carla Bagnoli, ‘Constructivism in Metaethics’, The Stanford Encyclopedia of Philosophy (2011), E.N. Zalta (ed.), available: <http://plato.stanford.edu/archives/win2011/entries/constructivism-metaethics/>, §2.3.

  57. 57

    I do not wish to suggest that this relativism is necessarily fatal for radical Humean constructivism. For instance, proponents might claim that a fairly stable set of human desires and judgements across most of humanity would lead to a de facto common set of norms to live by, sufficient to defuse the relativism worry. I shall not pursue this kind of response here, but simply wish to note that this would still represent a significant departure from the forms of objectivity promised by Kantian constructivism.

  58. 58

    On the political history of the term, see Martin Ostwald, Autonomia: Its Genesis and Early History (Chico, CA; Scholars Press, 1982).

  59. 59

    For the attribution of autonomy, see lines 820–2. Earlier, at line 453, Antigone emphasises that the divine law she obeyed is eternal. Sophocles, Antigone, The Women of Trachis, Philoctetes, Oedipus at Colonus, ed. and trans. Hugh Lloyd-Jones (Cambridge, MA: Harvard University Press, 1994), p.45. For a philosophical account of autonomy and reflexivity in the play, see David McNeill, ‘Antigone's Autonomy’, Inquiry 54: 5 (2011), pp. 411441.

  60. 60

    Epictetus, Discourses and Selected Writings, ed. and trans. Robert Dobbin (London: Penguin, 2008), IV.i.27 and IV.i.56 (see also I.xii.8–12); and Dio Chrysostom, ‘On Freedom’, Discourses LXI-LXXX, ed. and trans. H. Lamar Crosby (Cambridge, MA: Harvard University Press, 1951), pp. 313330.

  61. 61

    On this argument, see John Cooper, ‘Stoic Autonomy’, Social Philosophy and Policy 20:2 (2003), pp. 129.

  62. 62

    Romans, 2:14f.

  63. 63

    Thomas Aquinas, ad Romanos, §215f.

  64. 64

    Aquinas, ad Romanos, §217. For further discussion, see T.H. Irwin, ‘Continuity in the History of Autonomy’, Inquiry 54:5 (2011), pp. 447448.

  65. 65

    On the Leibnizian background to Wolff's account, see David Forman, ‘Free Will and the Freedom of the Sage in Leibniz and the Stoics’, History of Philosophy Quarterly 25:3 (2008), pp. 203219.

  66. 66

    Christian Wolff, ‘Reasonable Thoughts About the Actions of Men, for the Promotion of Their Happiness’, Moral Philosophy from Montaigne to Kant, ed. and trans. Jerome Schneewind (Cambridge: Cambridge University Press, 2003), p. 336.

  67. 67

    Wolff, ‘Reasonable Thoughts About the Actions of Men’, p. 334. This claim is redolent of Jürgen Habermas’ remark on the ‘the non-coercive coercion of the better argument’ in his Moralbewusstsein und Kommunikatives Handeln (Frankfurt am Main: Suhrkamp, 1983), p. 132.

  68. 68

    Wolff, ‘Reasonable Thoughts About the Actions of Men’, p. 337.

  69. 69

    Wolff, ‘Reasonable Thoughts About the Actions of Men’, p. 336.

  70. 70

    For Wolff, obligations arise from having motives attached to the obligated actions. The natural law is obligatory because ‘nature has connected motives with men's inherently good and bad actions’ (ibid, p. 335). This motivational analysis of obligatoriness opens Wolffian perfectionism to the well-known Kantian criticism that it is a heteronomous ethics unable to explain the categorical necessity of moral norms. However, realist accounts of self-legislation can reject a motivational analysis of obligatoriness. On the entanglement of motivational and normative elements in perfectionist and natural law accounts of obligation, see J. B. Schneewind, ‘Kant and Natural Law Ethics’, Ethics 104:1 (1993), pp. 5374. The importance of obligation (as opposed to other modalities of normativity) in relation to self-legislation is explored in Robert Stern, Understanding Moral Obligation: Kant, Hegel, Kierkegaard (Cambridge: Cambridge University Press, 2012).

  71. 71

    The Metaphysics of Morals contains the most well-known discussion. Kant, Practical Philosophy, p. 381, Ak 6:227. For comprehensive discussion and references, see Kain, ‘Self-Legislation in Kant's Moral Philosophy’. In contrast (and to my mind erroneously), Korsgaard claims, “In both Kant's version and mine the subject is unequivocally the author of the law” (The Sources of Normativity, p. 236).

  72. 72

    For a Schillerian approach to autonomy that develops these ideas, see Tom O'Shea, ‘Autonomy and Orthonomy’, Journal of Moral Philosophy (forthcoming), §8.

  73. 73

    I would like to thank Fabian Freyenhagen, Wayne Martin, Robert Stern, Christopher Hookway, Sonny Elizondo, an anonymous referee for this journal, and audiences at the University of Sheffield, the University of Essex, the Pacific APA, and the Wivenhoe Philosophy Circle for their comments on earlier versions of the article. Support was provided by the Arts and Humanities Research Council (grant AH/H001301/1).

Ancillary