1 Introduction

In the Groundwork, Kant regards the autonomy of the will as “the supreme principle of morality”. (GMS 4:440)Footnote 1 In the Critique of Practical Reason, he regards it as “the sole principle of all moral laws and of the duties conforming to them”. (KpV 5:33) Nevertheless, this impressively sounding principle gradually disappears from Kant’s writings during the 1790s, and there is not much in the literature to explain why. Pauline Kleingeld’s purpose, in the two articles I consider here, is to address this lacuna and to show that there are good philosophical reasons this principle drops out of the Kantian picture.

My aim is to formulate several reservations about Kleingeld’s argument; I do not claim to show that Kleingeld’s account is wrong, even if my reservations turn out to be apposite; I do, however, claim that, even if Kleingeld were right and Kant had abandoned the Principle of Autonomy, he would not have needed to do so.

The structure of the paper is quite straightforward: I start, in the next section, with a comprehensive presentation of Kleingeld argument; the aim is to introduce, as accurately as possible, the main elements of her position. I then formulate several reservations about her account, before ending with a brief conclusion.

2 Kleingeld’s View of the Formula of Autonomy and its Disappearance

That autonomy is a crucial part of Kant’s moral philosophy is the object of a rare broad consensus both among Kant commentators and more generally among those familiar with ‘Kantian’ philosophy. On Kleingeld’s account, one reason why we need to reconsider Kant’s account of autonomy in the Groundwork and the Second Critique is that it invites “voluntarist misunderstandings”, according to which “one puts oneself under moral obligation by a contingent act of will”; this implies that “one can repeal the law or enact a different law if one wants to”. (Kleingeld, 2018a: 159) Two questions are invited by such misreadings: (1) what ‘autonomy’ means and (2) how Kant could have considered the idea of autonomy as apt for expressing the moral law.

To begin with the second question, we can note that one assumption with which Kleingeld starts is an interpretation of Kant’s claim that the different formulae of the moral law involve the use of “a certain analogy” and that they are different ways in which we can “represent the principle of morality”. (GMS 4: 436, emphasis added by PK; also GMS 4: 437) The formulas of the moral law “bring an idea of reason closer to intuition” and, in this way, closer to feeling. (GMS 4: 436) On Kleingeld’s account, Kant introduces the Formula of Autonomy with the help of an analogy with political legislation. In imperatival form, the Formula reads as follows: “act as if your maxim were to serve at the same time as a universal law (of all rational beings)”. (Kleingeld, 2018a: 160; GMS 4: 438).

Two further questions Kleingeld thinks are raised at this point by a reconsideration of Kant’s idea of autonomy. First, we can ask (3) whether the Formula of Autonomy is supposed to refer to a formulation of the Categorical Imperative (with capital letters, referring to the imperatival form of the Moral Law) or of a categorical imperative (with small letters, and referring to the maxims which pass the test of the Categorical Imperative and prove themselves in this way to be valid commands – for instance, a prohibition of lying). Secondly, we can consider (4) whether Kant can legitimately refer to a principle of “autonomy” at all, given that the term does not appear in the formulation of the principle, which does not even include an explicit reference to the self.

Kant’s main political writings are only published in the 1790s. Hence, the standard picture of Kant’s use of ‘autonomy’ in the Groundwork is that Kant developed this notion on the basis of Jean-Jacques Rousseau’s political conception of autonomy from the Social Contract. Kant was familiar with this work by Rousseau. Kleingeld notes, however, that the standard view overlooks the fact that Kant was lecturing on legal and political theory at the time when he was writing the Groundwork. Moreover, the very first documented public reference to the idea of moral autonomy is identified in the introduction to these lectures (the Feyerabend Lectures from 1784). Finally, there are similarities between Kant’s account of moral autonomy and his account of political legislation. (Kleingeld, 2018a: 161–2).Footnote 2

As mentioned, Kant claims that the different formulas of the moral law involve the use of analogies and that they are different ways in which we can represent the principle of morality. Kleingeld notes that, for Kant, analogies play a very important role as a means for the representation of ideas; whereas empirical concepts can be illustrated by examples and a priori concepts of the understanding can be schematised, the ideas of reason need analogies for representation. Analogies are not simply decorative, but have an important role in the clarification of ideas.

By an analogy, Kant does not mean “an imperfect similarity of two things”, but “a perfect similarity of two relations between entirely dissimilar things”. (Prol 4: 357 – emphasis added by PK, who quotes this in 2018a: 163) For example, in drawing an analogy for the idea of God, Kant claims that there is a perfect similarity between the relation between God and the sensible world, on the one hand, and, on the other, between a watch and its maker (or between a ship and its builder or a regiment and its commander). The point of the analogy is to argue that we can regard the sensible world as if ordered by a supreme being and to proceed, when investigating nature, as if nature had a rational order. Concerning the Formula of Autonomy, Kleingeld notes, Kant says that we are to regard ourselves as giving universal laws through our maxims or we are to act as if we were giving such laws.

The discussion so far suggests that, in stating the Formula of Autonomy, Kant offers a representation of the moral law, in order to bring this law closer to intuition and that he is doing this through an analogy, which, Kleingeld argues, draws a similarity of relations with elements of Kant’s account of political legislation in the Feyerabend lectures. Kleingeld’s claim is that

the relation between the Categorical Imperative and the ‘universal law’ mentioned in the Principle of Autonomy […] parallels the relation between the a priori constitutional law of the state and positive state laws. (Kleingeld, 2018a: 164)

This provides an answer to question (3) above: like the Categorical Imperative, the Principle of Autonomy is a test for the maxims which are candidates for the status of categorical imperatives.

In the Feyerabend lectures, Kleingeld notes, Kant argues that the main normative principle for achieving justice is the a priori idea of an original contract – a just state is a state originated in the agreement of the people to subject themselves to common legislation. (NF 27: 1382) This implies that any positive legislation ought to be consistent with this idea of an original contract and, hence, ought to be such that it could have proceeded from the agreement of the entire people. (Kleingeld, 2018a: 164–5) The criterion of rightness for positive laws is whether they “could have arisen from the agreement of all”. (NF 27: 1382) The criterion, Kleingeld notes, is neither actual nor hypothetical, but modal agreementFootnote 3:

The normative constitutional principle requires neither that the people actually consent to the laws nor that they would consent if asked. It merely requires that the people could consent to them. (Kleingeld, 2018a: 165 – emphasis in the original)

Hence, Kleingeld notes, the criterion of rightness for positive law is whether the law is genuinely universal.Footnote 4 For instance, a ruler who imposes taxes on merchants, but makes exception for his favourites fails to legislate justly. The law of taxing some citizens and making exceptions for the others is not universal – the exception is made on the basis of a particular interest, which the ruler happens to have.

Hence, according to this reading, given the analogy with political legislation, the Principle of Autonomy is a formulation of the Categorical Imperative, which corresponds, on the political level, to the constitutional or fundamental law. This constitutional law demands that the moral agent act as if he were a legislator of universal laws, which are conceived of as analogous to positive state laws. There are three main parallels between the moral domain and political domains that Kleingeld focuses on.

First, Kant regards the constitutional principle and the Categorical Imperative as a priori principles of pure reason. The parallel, Kleingeld notes, is even clearer in the Second Critique, where he talks about the Categorical Imperative by using the term ‘Grundgesetz’ – that is, constitution or constitutional law. (KpV 5: 30, 31, 43, referred to in Kleingeld, 2018a: 166) Secondly, Kant regards both the constitutional principle and the Categorical Imperative as guiding in the justification of moral and political principles. Hence, the laws given through one’s maxims are the analogical counterparts of positive state laws. Thirdly, the normative content of the a priori constitutional principle is mirrored in its analogical moral counterpart – in both cases, the candidate maxim or the proposed state law ought to be able to serve as a universal law.Footnote 5

Kleingeld emphasises that Kant’s formulation of the moral imperative indicates that we are not actually expected to legislate through our maxims: “Act as if your maxim were to serve at the same time as a universal law”. (GMS 4: 438) Thus, Kleingeld explains by reference to the Kantian “act as if” expression that “[t]he moral agent is not required literally to give laws but rather to determine the permissibility of his maxims by using the model for evaluating the justice of state laws”. (Kleingeld, 2018a: 12)Footnote 6 The model for evaluating the justice of state laws is similar to that for evaluating moral principles and both refer to the notion of contradiction. Kant says that contradictions will emerge when the legislator’s “private choice” is used to formulate laws, which are supposed to be the object of common agreement. Similarly, Kant thinks that moral maxims may fail the test of the Categorical Imperative when they lead to a contradiction.

To explain the sense in which Kant talks about autonomy in the Groundwork, Kleingeld starts from Kant’s discussion of political legislation in the Feyerabend Lectures. In particular, she starts from Kant’s claims that the people are sovereign and that the highest legislation belongs to those who obey the law. He does suggest that actual legislation be undertaken by the people’s representatives, but he does not argue that citizens ought to have the right to vote and legislate indirectly or directly. He does allow for political systems which include actual legislation by their citizens, but he does not think actual legislation by the citizens would be a necessary condition for justice. (Kleingeld, 2018a: 169).

On Kleingeld’s reading, Kant understands people’s sovereignty as requiring only that the highest legislator who represents them give laws that the entire people could agree to. This is to be distinguished from what laws citizens actually agree with or would agree with, if given the opportunity. Thus, Kant is reported to claim that “[t]he laws of a despot can be just, when they have been made such that they could have been made by the entire people”. (NF 27: 1382) Hence, “the idea of self-legislation, or the idea of the autonomy of the people, functions as a normative criterion”. (Kleingeld, 2018a: 170).Footnote 7

This discussion of autonomy starts to answers the questions (1), (2) and (4) above. Kant takes autonomy to refer to self-legislation, but without voluntarist connotations. Those standards (whether positive laws or maxims) are considered legitimate, which could universally be agreed to. Hence, although more details are to be added, it seems clear that Kant legitimately calls this formula of the moral law the ‘Autonomy Formulation’, despite the fact that neither ‘self-legislation’ nor even the word ‘self’ figures explicitly in it. What is important here is that every agent can see herself as the source of the laws.

Next, Kleingeld turns to the Formula of Autonomy in the Groundwork and the first feature she discusses is the “counterfactual” character of Kant’s criterion. On Kant’s account, we are to regard ourselves as if we legislated universally through our maxims; yet, this does not imply that rational beings must give their consent or that we need to take into consideration the actual attitudes of others or to engage in joint deliberation. Hence, the agent who regards himself as legislating universal laws through his maxims seems to be the moral analogue of the political legislator.

One important aspect this comparison highlights is that the moral agent is to regard himself as giving universal law – that is, giving a law to all, not simply to oneself, as often understood in the literature. In the same way in which a political legislator gives law to the entire people, rather than primarily to himself, the moral agent gives law to the entire ethical community. To be sure, genuinely universal law also applies to the legislator himself, and Kant, on Kleingeld’s reading, is quite clear in this respect. All human beings should regard themselves not merely as lawgiving, but also as subject to the laws they regard themselves as giving.

Hence, on Kleingeld’s reading, ‘autonomy’ means being subject to laws that are one’s own, as opposed to laws one is given by another. Hence, in the Formula of Autonomy, one is to regard oneself as giving universal laws and as being subject to these laws. In the Groundwork, Kant notes:

In accordance with this principle all maxims are rejected that cannot coexist with the will’s own universal legislation. The will is thus not merely subject to the law, but subject in such a way that it must also be viewed as self-legislating [selbstgesetzgebend] and precisely for that reason subject to the law in the first place (of which it can regard itself as author [Urheber]). (GMS 4: 431 as cited by Kleingeld, 2018a: 173)Footnote 8

Hence, for Kant, autonomy means living under one’s own laws. The Formula of Autonomy enjoins agents to regard themselves as giving universal laws, to which they themselves are subject. This does not mean that agents actually give moral laws. The analogy with politico-juridical legislation shows that Kant has in mind a test for the permissibility of one’s maxims. Hence, as already mentioned, the idea of autonomy does not have voluntarist implications. The content or validity of moral laws is not dependent on an act of the agent or on her actual consent. Hence, the agent is not at liberty to rescind moral laws either. Questions (1)-(4) can now be seen as completely answered.

Thus, in answer to question (1) (what ‘autonomy’ means for Kant), we have seen that this notion refers to a situation where the agent is being subject to laws that are one’s own, not given by another. In response to question (2) (how Kant could have considered the idea of autonomy as apt for expressing the moral law), we can note that the crucial aspect is that the Formula of Autonomy enjoins us to regard ourselves as giving universal laws to which we are subject. Question (3) (whether the Formula of Autonomy has the status of a maxim or of a principle which provides a test for maxims) led us to the conclusion that the Formula of Autonomy has the status of the Categorical Imperative and applies to maxims. Finally, in reply to question (4) (whether Kant is justified in talking about the Formula of Autonomy, when this principle does not make any reference to autonomy or to a reflexive ‘self’), we can refer directly to the answers to questions (1) and (2). Thus, the Formula of Autonomy enjoins the agent to regard herself as autonomous, as the subject of laws that are her own.

Next, Kleingeld considers also another aspect of Kant’s discussion of autonomy in the Groundwork. Thus, on her account, Kant also talks about autonomy as a property of the will. More exactly, it is “the will’s property of being a law to itself”. (GMS 4: 440) This means that the will has the property of autonomy and it is a law to itself, rather than analogically regarding itself as such. Kleingeld’s claim is that this sense of autonomy as a property of the will is not discontinuous with the sense established through the analogy with political legislation; rather, the former follows from the latter.Footnote 9

The Categorical Imperative, as an a priori principle of reason, Kleingeld notes, is also a principle of the will, since Kant identifies the will with pure practical reason. The Categorical Imperative is a principle of the will both in the sense that it belongs to the will or practical reason and in the sense that the will or practical reason are subject to the Categorical Imperative and, hence, to its own law (and the imperatives which are justified by the Categorical Imperative). As a result, Kleingeld concludes that

Kant can describe the will as having the property of autonomy in the sense of its being subject to its own laws, without this implying that the obligatory force of moral laws (and the corresponding duties) derives from a contingent act. (Kleingeld, 2018a: 174 – emphasis in the original)

Now that the notion of autonomy is clearer, the question raised by Kleingeld concerns the virtual disappearance of the notion of autonomy of the Formula of Autonomy from Kant’s later writings, in particular from the Metaphysics of Morals. Kleingeld’s answer starts from the analogy with Kant’s political theory in the Feyerabend lectures; her claim is that, in his later writings, this political theory changes and the analogy with autonomy is no longer apt. During the early 1790s, Kant no longer thinks that genuine universality suffices for the justice of state laws. He adds the requirement that the laws be given by the citizenry. Hence, Kleingeld concludes,

the idea of “legislating through one’s maxims” was no longer suitable as an analogy with which to express the moral principle. This explains why Kant no longer mentions the Principle of Autonomy. (Kleingeld, 2018b: 62)

This seems to suggest that, with the change in his political theory, Kant changed his view of political legislation and the link to moral legislation was no longer analogical. Legislating through one’s maxims still meant to determine whether particular maxims could be adopted by all. Yet, political legislation now required more – in addition to this modal test of consent, Kant introduced a condition of actual consent. The analogy between Kant’s moral and political accounts of legislation disappeared and, with it, also the Formula of Autonomy.

One aspect of Kleingeld’s reading presented so far is worth emphasising. On her account, there is an important difference between two ways of looking at autonomy. First, we can regard autonomy as the will’s considering itself as giving laws to itself and only mediately, through the idea of the will as pure practical reason, implicitly to all rational agents. There is, secondly, the view of autonomy as the will’s conceiving of itself counterfactually as legislating to all (including itself). The difference seems to be important because Kant is using a political analogy, and it would be a misrepresentation to say that the role of a political legislator is to give laws to himself:

[O]ne would misdescribe the will’s imagined objective if one were to say that its aim is to give law to itself; its imagined objective, as Kant describes it, is to give universal laws – laws that apply to all rational beings. (Kleingeld, 2018b: 66)

Hence, on Kleingeld’s reading, if the will is to consider itself as giving law to itself, then it misses its appropriate imagined objective, which is to give laws for all members of the ethical community. Giving law to oneself suggests the idea of a primarily self-addressing act, which misrepresents autonomy. The importance of this distinction, Kleingeld notes, is given by the implications for consent: if the will is viewed as legislating to itself, then its consent is implied. If the will is viewed as legislating to all rational beings, then this raises questions concerning the imagined consent. As we have seen, on Kleingeld’s account, for moral autonomy, Kant endorses a modal consent, rather than a hypothetical or actual one. A view of autonomy as legislation to oneself not only is actually associated with a different type of consent than that required by Kant’s account, but also makes impossible an association with another kind of consent, since it presupposes the actual one.

Now, according to Kleingeld, Kant’s view of political legislation changes in the 1790s. She notes that Kant introduces the normative requirement that citizens actually consent to legislation. He no longer thinks it suffices that the people could agree to a law, but now adds the requirement that the citizens also do agree to it through their elected representatives in parliament. There is, therefore, a move from a normative criterion which requires modal consent, to one which requires actual consent.

In the Metaphysics of Morals, Kleingeld claims, Kant regards modal consent as only a necessary condition for the just character of laws, and adds the additional condition of an actual consent. Kant views the constitution of a “pure republic” as the “only constitution that accords with right” (MS 6: 340) and he regards a pure republic as a “representative system of the people, in order to provide it with its rights, in its name, by all the state citizens united, by means of its delegates (deputies)”. (MS 6: 340) Moreover, Kant thinks now that the legislative authority belongs to the united will of the people, who, insofar as they are united for the purpose of legislation, are called citizens. He regards the capacity to vote as the only qualification for being a citizen.Footnote 10

Kant also specifies that the freedom of the citizen implies “obeying no other law than that to which he has given his consent”. (MS 6: 314) Kleingeld takes this to suggest that “right requires that the citizens do consent (not individually but collectively, in parliament) to the laws to which they are subject”. (2018b: 74 – emphasis in the original) Moreover, Kant specifies other rights for citizens, apart from the right to vote, such as the right to manage the state, to organise it or to cooperate for introducing certain laws. Although Kant is still committed to a priori normative constraints on political legislation, Kleingeld notes, the crucial difference for her purpose is the addition that the laws be given by the citizens themselves, through elected representatives.

Kleingeld’s conclusion is that:

Kant’s earlier analogy with political legislation became unsuitable as a formal model for assessing the moral permissibility of maxims. Using the old analogy would have been awkward. Kant came to reject the idea that legislation can be fully just without the actual consent of the citizens, and it would have been strange for him to articulate the principle of morality in terms of a political model he had discarded. (Kleingeld, 2018b: 75)

The suggestion then seems to be the following. The moral law is an idea of reason, which needs analogies to be brought closer to intuition. One way to bring the moral law closer to intuition is by an analogy with political legislation, an analogy which brings forward aspects of the moral law Kant captures through the Principle of Autonomy. Call the view of political legislation involved in the analogy PL1. Yet, once Kant’s view of political legislation changes, he can no longer endorse PL1, and, hence, he can no longer endorse the analogy with the Principle of Autonomy.

On Kleingeld’s reading, the normative principles governing moral and political legislation are no longer fully analogous. In both domains, the criterion of universality remains in force, but Kant adds in the political domain the requirement that laws be adopted by citizens, a requirement which does not have a parallel in the ethical domain. Hence, the normative constraints on political legislation are no longer suitable for articulating the moral constraints on one’s maxims of action.

Kleingeld adds that Kant continues to use political analogies in his ethics. She attempts to determine “how, if at all, we might redescribe the old analogy in terms of Kant’s new political philosophy”. (Kleingeld, 2018b: 76) In Kant’s new political philosophy, the old question of the universality of political laws is still raised. It is raised, Kleingeld claims, when “citizens ask themselves whether a legislative proposal is a real candidate for legislation – that is, whether a proposed law meets the formal requirement of being genuinely general”. (Kleingeld, 2018b: 76)Footnote 11 Hence, the question is whether the law qualifies for political legislation, rather than being directed at serving merely private interests or privileging certain groups on some contingent considerations (for instance, hereditary). Kleingeld thinks that this is how Kant reformulates the procedure for testing the ethical permissibility of one’s maxims – one should ask whether one’s maxim simultaneously “qualifies” for general legislation. (MS 6: 225, 226, 389, 393, 451, cf. 214). Hence, Kleingeld concludes:

the moral imperative no longer directs agents to act as if they were giving universal laws through their maxims; instead, it directs them to act on maxims that could become universal laws (and simultaneously serve as their own maxims). (Kleingeld, 2018b: 76 – emphasis in the original)

The claim seems to be, therefore, that the old analogy with political legislation or the giving of political laws is replaced with a new analogy, where maxims which could become universal laws qualify as ethical principles or imperatives. This, Kleingeld acknowledges, is not an entirely new analogy and Kant does not elaborate the idea in detail. But the suggestion is that, although the notion that one’s maxim should be “fit” or “appropriate” for universal legislation is already implicit in the old legislation analogy in the Groundwork and the Second Critique, it becomes the leading idea in the Metaphysics of Morals.

3 Some Reservations about Kleingeld’s Reading

I find Kleingeld’s discussion of autonomy very illuminating in its clarification of the sense in which Kant talks about the Formula of Autonomy and autonomy as a property of the will. My comments, however, will not focus on these aspects of her papers, but on her account of why the Formula of Autonomy is mentioned less frequently by Kant after the Groundwork, and not mentioned at all in the 1790s writings.Footnote 12 The fundamental part of her account consists of the observation that Kant changes his view of political legislation in the 1790s and that the Formula of Autonomy relies on an analogy with political legislation; if Kant can no longer support the needed analogy for the Formula of Autonomy, then it should not be surprising that his references to it decrease in frequency. My question, however, is whether this change in Kant’s view of political legislation is sufficiently significant to lead to what would effectively be a rejection of this formulation of the Categorical Imperative. My strategy will be to show that the differences introduced by Kant’s new conception of political legislation do not change significantly the elements needed for the analogy on which the Formula of Autonomy allegedly relies.

First, however, a terminological clarification: throughout the two papers considered here, Kleingeld draws a distinction between the moral and the political, for instance, between political and moral legislation, between political and moral communities or between political and moral agents. In what follows, I will employ a different distinction. This is presented by Kant in the “Introduction to the metaphysics of morals”, the section “On the division of a metaphysics of morals”. It is a distinction between ethics and political-juridical philosophy, which Kant takes to be parts of moral philosophy or, more precisely, of a metaphysics of morals. (MS 6: 218–9) Accordingly, I will distinguish between ethical and political agents, communities, legislations.

Consider Kant’s idea of lawgiving; in Section IV, “On the Division of a Metaphysics of Morals”, of the Metaphysics of Morals, Kant says:

In all lawgiving (whether it prescribes internal or external actions, and whether it prescribes them a priori by reason alone or by the choice of another) there are two elements: first, a law, which represents an action that is to be done as objectively necessary, that is, which makes the action a duty; and second, an incentive, which connects a ground for determining choice to this action subjectively with the representation of the law. Hence, the second element is this: that the law makes duty the incentive. By the first, the action is represented as a duty, and this is a merely theoretical cognition of a possible determination of choice, that is, of practical rules. By the second, the obligation so to act is connected in the subject with a ground for determining choice generally. (MS 6: 218)

Kant says explicitly that each lawgiving [Gesetzgebung] has two parts: a law and an incentive. The law represents an action, which, as represented by the law, is objectively necessary and, hence, a duty. The incentive connects a ground for determining choice with the representation of the law and the connection takes place subjectively. Without the incentive, the law presents a theoretical cognition of a possible determination of choice (practical rule). That is, without a motive to perform the action represented by the law, the law formulates a duty, which presents itself to me as a possible action and, hence, as a possible rule of action.

According to Kant, depending on the kind of law we have, we can end up with the following types of lawgiving:

All lawgiving can therefore be distinguished with respect to the incentive (even if it agrees with another kind with respect to the action that it makes a duty, e.g., these actions might in all cases be external). That lawgiving which makes an action a duty and also makes this duty the incentive is ethical. But that lawgiving which does not include the incentive of duty in the law and so admits an incentive other than the idea of duty itself is juridical. It is clear that in the latter case this incentive which is something other than the idea of duty must be drawn from pathological determining grounds of choice, inclinations and aversions, and, among these, from aversions. (MS 6: 218-9)

Kant’s point here is quite clear: if a law, as an element of lawgiving, is acknowledged by an agent as objectively necessary, then it can be given as a law to be followed either solely in virtue of its objective validity or not solely in virtue of its objective validity or necessity (for instance, in virtue of an aversion, presumably to punishment). In the former case, that is, when the law is given as to be followed because this would be the right thing to do, Kant identifies an ethical type of lawgiving. In the latter case, that is, when the norm is not given as to be followed solely with the ethical motivation (but, say, in order to avoid punishment), Kant talks about juridical lawgiving. This is one instance of the distinction between the ethical and the political, which I am going to use in this paper.Footnote 13

Let me formulate now some of the reservations I have about Kleingeld’s account of why the Principle of Autonomy is no longer mentioned by Kant, beginning with the 1790s. First, on Kleingeld’s reading, the fundamental change in Kant’s account of political legislation is the following: according to the old account of political legislation, “a highest legislator who represents them ought to give only laws that the entire people could agree to (or could have agreed to)”. (Kleingeld, 2018a: 169 – emphasis in the original) By contrast, for Kleingeld, on Kant’s later account of political legislation,

[h]e no longer regards it as sufficient that a law is truly general and could be adopted by the people as a whole. He now regards this as a merely necessary condition, adding the further requirement that legislation must result from actual consent by the citizens, via their elected representatives in parliament. (Kleingeld, 2018b: 173)

As already mentioned, the emphasis placed by Kleingeld in both her papers is on a change from a requirement for a merely modal consent to a requirement which, in addition to a necessary modal consent, introduces also as necessary an actual consent. But, as can be seen from these quotations, what we have is a slightly different change: in both cases, political laws are actually consented to by the citizens’ representatives; what changes is who these representatives are. After all, the highest legislator needs to actually consent to the laws enacted for the entire people, if he is to give those laws, in the same way in which the representatives of the people on the new model need to consent to the laws they will give. Laws which are truly general and could be adopted by the people as a whole are not fully justified on the new conception of political legislation. Yet, they would not have been fully justified on the old conception either. These laws need, as it were, to result from the actual consent by the citizens, via their representative, just as they do on the new account of political legislation.

A second aspect which I find puzzling is the following: on Kleingeld’s account, for the Formula of Autonomy, the analogy with political legislation suggests that, for the ethical domain, the ethical agent, similarly or analogously to the political legislator, will legislate for all members of the community; yet, the change in Kant’s account of political legislation is not modifying this aspect either. The elected representatives of the citizens are still supposed to legislate for all members of the community. After all, as acknowledged by Kleingeld, Kant’s new view of political legislation preserves the status of an a priori normative principle for the idea of the original contract. Hence, the requirement that political decisions be made in such a way that they could be adopted by the people as a whole still applies.

Kleingeld suggests that crucial for autonomy is the notion that “the will is the source of legislation (that is, legislation by the will itself) […and not being] subject to laws given by another”. (Kleingeld, 2018b: 67–8) As mentioned, she draws a distinction between autonomy, as the property of the will, and the Formula of Autonomy, as the specific formulation of the Categorical Imperative, which is obtained through the analogy with political legislation. Kant’s account of autonomy, as a property of the will, Kleingeld writes, does not change, when Kant’s account of political legislation changes in the 1790s. But does the new account of political legislation challenge the conception of autonomy captured by the Formula of Autonomy? We have seen that the Formula of Autonomy requires us to regard ourselves as subject to laws that are our own. If this aspect of autonomy was there in the case of the old view of political legislation, however, then it will appear even more clearly in the context of the new view of political legislation.

Thus, on the old view of political legislation, it is the highest legislator who makes the laws for all members of the political community, including for himself. It is true, these laws need to pass the test of modal consent – it should be possible for all members of the political community to consent to these laws. Yet, any other member of the political community, apart from the highest legislator, will see these laws as the laws of another (namely, the laws of the legislator), although, given that they meet the condition of modal consent, they could also be their own. By contrast, on Kant’s new account of political legislation, members of the community, although they may not play also the role of citizens’ representatives, can more easily see the laws passed by representatives as their own, since they elected these representatives.

Thirdly, let us consider what seems to be a clear proof that Kant abandons the Formula of Autonomy, because it relies on an account of political legislation, which he no longer supports. Thus, in a section entitled “The Obsolescence of the Legislation Analogy”, Kleingeld offers the following account of Kant’s discussion of the ethical community in the third section of Division 1 of Religion’s third part. Kant notes that, in the case of a republic or juridical commonwealth, the people as a whole is the legislator; in the case of a universal republic according to laws or virtue or the ethical commonwealth, the people itself cannot be regarded as legislating. The legislator must be someone other than the people, namely, God. Kleingeld concludes:

Kant here seems to indicate that it is impossible to update the Groundwork idea of the “realm of ends” in the form of a moral analog of his new political ideal of a republic in which the united citizens themselves legislate.

In sum, Kant’s political theory changed in a way that rendered the analogy with political legislation unsuitable. The old legislation analogy became obsolete. (Kleingeld, 2018b: 75)

Kleingeld argument, it seems to me, is the following: in the Groundwork, where Kant discusses the Formula of Autonomy, he relies on an analogy with political legislation and enjoins each ethical agent, as a member of the ethical community, to act as if their maxims were to serve at the same time as a universal laws for all members of the ethical community; in Religion within the boundaries of mere reason (1793), he relies on the new account of political legislation, one which no longer sustains the previous analogy and the view of ethical legislation from the Groundwork; consequently, Kant takes the legislator to be God and he no longer seems to admit that every ethical agent should act as if they were giving laws to all the other members of the ethical community. Political legislation is viewed by Kant differently now, in a way which makes unsuitable an analogy with ethical legislation; ethical legislation is now conceived of on the basis of an analogy with divine legislation.

Consider now Kant’s discussion in the Religion; he thinks that, in the case of a political (or juridical) community, “the lawgiver (of constitutional laws)” must be “the mass of people joining in a union”; this is “because legislation proceeds from the principle of limiting the freedom of each to the conditions under which it can coexist with the freedom of everyone else, in conformity with a universal law”, and what is established in this way is “an external legal constraint”; by contrast, for an ethical community, “all the laws are exclusively designed to promote the morality of actions (which is something internal, and hence cannot be subject to public human laws”; Kant adds that public laws “are directed to the legality of actions, which is visible to the eye” and “in this they constitute a juridical community”; for an ethical community “(inner) morality […] alone is at issue”. (RGV 6:98–9 – emphases in the original).

Two notes are worth adding here. First, the distinction between morality and legality is already formulated in the Groundwork and Critique of Practical Reason along the same lines (for instance, KPV 5:71–2 and see also below) Secondly, in Religion, Kant does not simply claim that ethical laws are legislated by God; on the contrary, he notes that, if ethical laws are thought as proceeding “originally merely” from the will of God, “as statutes that would not be binding without his prior sanction”, then “they would not be ethical laws, […] but an externally enforceable legal duty”; instead, God can be seen as supreme lawgiver of an ethical community, if “all true duties, hence also the ethical, must be represented as at the same time his commands”. (RGV 6:99 – emphases in original).

My claim is that Kant’s discussion in Religion is in fact an implication of the distinction between ethical and juridical lawgiving, a distinction introduced in more detail at the beginning of this section, and present in Kant’s work in the form of the distinction between legality and morality already in the Groundwork and the second Critique. Hence, the discussion does not bear on the issue discussed in Kleingeld’s two papers and examined here.

In Religion, Kant focuses on the incentive specific for ethical lawgiving. As we have seen, ethical lawgiving makes an action a duty and makes that duty the incentive with which the action is to be performed. Such an action has morality. Since ethical lawgiving presupposes this restriction on the incentive with which the action is to be performed (a restriction which is missing in the case of juridical lawgiving), ethical laws cannot be given either by a human legislator (a highest legislator) or by a groups of human legislators (the citizens’ representatives) or simply by a divine legislator; in all these cases, laws may be formulated, but the incentive with which each person is to follow those laws is the result of each of these persons’ determining themselves to act out of the acknowledged duty to follow the laws.

The reason why Kant does not simply regard each person as a legislator and suggests that God is the supreme legislator of the ethical community has to do with a complexity that Kant introduces in fact in the Groundwork. Kant mentions there that we cannot be sure of the incentive with which other persons and even we ourselves perform actions:

[I]f we attend to our experience of the behaviour of human beings we meet frequent and, as we ourselves concede, just complaints that no reliable example can be cited of the disposition to act from pure duty; that, though much may be done that conforms with what duty commands, still it is always doubtful whether it is done from duty and thus has a moral worth. […]

In fact, it is absolutely impossible by means of experience to make out with complete certainty a single case in which the maxim of an action that otherwise conforms with duty did rest solely on moral grounds and on the representation of one’s duty. (GMS 4:406-7)

This is a well-known problem, sometimes called the Opacity Thesis, whose significance is still debated in the literature. For the purpose of this paper, suffice it to say that, in Religion, Kant takes God to be the legislator of the ethical community, because this legislator, whoever they would be,

must also be one who knows the heart, in order to penetrate to the most intimate parts of the dispositions of each and everyone and, as must be in every community, give to each according to the worth of his actions. (RGV 6: 99)

Yes, Kant notes, “this is the concept of God as a moral ruler of the world”. (RGV 6:99) The reason why Kant says that all true duties (including ethical ones) must be represented as at the same time God’s commands is that, in following these laws we should not simply act in accordance with them, but also because we acknowledge they are duties, actions which we ought to perform. Such an acknowledgement, however, depends also on the extent to which we actually act on them because they are duties. If I follow a law, say, because I want to avoid the feeling of guilt produced by non-compliance, then what I acknowledge, in fact, is the conditional value of the action, not its moral worth. Given that we can never be absolutely sure that we performed an action out of duty, the best we can do is to act in such a way that we can represent these duties as at the same time God’s commands, that is, commands potentially monitored by a being who has knowledge of the incentives with which we actually act.

The discussion concerning the Formula of Autonomy, however, is not about the incentive of ethical lawgiving, but the duty of the lawgiving – the law represented as to be followed. The condition of modal and actual consent is a condition for the laws formulated by the respective lawgivers. By contrast, in Religion, Kant talks about the incentive of ethical lawgiving. This is evident from the footnote at 6:99, where Kant mentions that, when something is recognised as duty, “even if it should be a duty imposed through the purely arbitrary will of a human lawgiver, obeying it is equally a divine command”. (RGV 6: 99n).

Hence, there is first the issue of the formulation of the law of lawgiving, and this can be done, for juridical lawgiving, by, say, the highest legislator or by citizens through their representatives in the parliament; for ethical lawgiving, it will be done by the ethical agent, who may take themselves to be a highest legislator or a representative of citizens. Secondly, there is the issue of the incentive with which duties are to be fulfilled; for all duties, whether specifically juridical or ethical, they can be performed with an ethical incentive (out of duty) and, in this case, they should also be representable at the same time as God’s commands. Nevertheless, juridical lawgiving does not impose a condition on the incentive with which juridical laws are observed, so here the issue of identifying the incentive for a particular action does not arise, and the condition of possible simultaneous representation as divine command is not relevant either.

Let me conclude this section with one final puzzle. Perhaps the crucial differences between Kant’s two accounts of political legislation are the following: on the earlier account, the highest legislator is not elected by the members of the community, whereas in the later, the representatives of the citizens are; moreover, on the earlier account, laws are given without deliberation, as it were, monologically; by contrast, on the later account, they are the result of deliberation or, in other words, the process is dialogic.Footnote 14 What is puzzling is that these differences do not seem to make the initial analogy in any way problematic or unsuitable for Kant’s purposes.

Both differences introduce complexities into Kant’s later account of political legislation, complexities which may seem to provide prima facie reasons why he would no longer see an analogy between this new account of political legislation and the Formula of Autonomy. Yet, if we look closer, it becomes clear that the analogy could easily function equally well.

Consider the first difference: the highest legislator is not elected by citizens, whereas the representatives – members of parliament – are. Why would the analogy with the ethical agent be appropriate in the first case, but not in the second or why would it be more appropriate in the first case than in the second? This is one possibility: when a person is in an ethically relevant situation, she must decide what the ethical thing to do is; her position seems closer to that of the highest legislator, rather than that of the elected representative; after all, she is not elected to represent somebody else.

Yet, the highest legislator is supposed to represent the members of the political community, so, in that respect, the relation of the ethical agent to the highest legislator is not different from her relation to the elected representative. It might be replied that the highest legislator is not elected, in the same way in which the ethical agent is not elected. Yet, of course, the person playing the role of the highest legislator must have gone through a process of being invested with the role of highest legislator, a process through which neither the elected representative nor the ethical agent had to go. There will certainly be similarities and differences between aspects of the persons or processes between which we draw an analogy. As Kleingeld herself notes, however, what is relevant, for Kant, in an analogy is not a specific set of such similarities, but the identical relations between specific aspects of those persons or processes.

For instance, the ethical agent, the highest legislator and the elected representatives regard themselves as legislating for all members of the relevant community by actually choosing from among the laws which could be consented to by all members of the community. These are laws to which they are also subject, and they are the source of these laws, they are not subject to laws given by others. Here, however, it might seem relevant to consider the second important difference between Kant’s two accounts of political legislation – the difference between their monological and dialogic characters.

Thus, it might be argued that the elected representative does not simply give laws to all members of the political community, but is involved in a process of deliberation in the parliament, a process which leads to laws actually consented to by the citizens’ representatives. This dialogic process of legislation is missing both in the case of the ethical agent and in the case of the highest legislator. This might seem to be a sufficiently important consideration for Kant to abandon the Formula of Autonomy, which regards ethical legislation as monological by analogy with an account of political legislation he no longer endorses and effectively rejects.

In reply to this, recall first the discussion of Kant’s argument in Religion: in the case of ethical lawgiving, unlike the case of juridical lawgiving, it is the ethical agent who will formulate the laws as if they were to serve as laws for all members of the ethical community. More importantly, however, Kant’s later conception of political legislation can be seen as a very good model for this process of formulation of ethical laws. After all, as Kant mentions in the Groundwork, the ethical agent does go through a deliberative process in which the will or other (alien) impulses may give laws and at the end of which a law will actually be consented to by the ethical agent. (GMS 4:444) To be sure, as in the case of a debate in the parliament, some decisions will confirm the positions of some parties in the deliberative process, whereas other decisions will be in favour of other parties. Sometimes the process will lead to the morally appropriate outcomes, whereas on other occasions, it will reach the wrong conclusions. This can happen for both an intrapersonal and an interpersonal deliberative process.

To be sure, the process of political legislation is dialogic in a way in which the ethical one is not. Yet, again, essential is not which set of similarities and differences can be established between the analogues, since the analogues are supposed to be relations, on Kant’s account, and, moreover, they are supposed to be identical relations. Yet, the relations which can be established between the ethical process of legislation and Kant’s earlier account of political legislation can also be established between the ethical process and Kant’s later account.

4 Conclusion

If Kleingeld’s account of why the Formula of Autonomy has gradually disappeared from Kant’s writings in the 1790s is not convincing, how can we explain this disappearance? I think the answer can be found in Kant’s idea that the various formulae of the Categorical Imperative are ways in which we can bring the Categorical Imperative closer to intuition. Kant suggests that there are various ways in which this can be done, depending on the aspect on which we focus. Hence, on possibility is that, after 1790, he focuses on different aspects than those relevant for the Formula of Autonomy.

As an illustration, consider the way in which, in the Doctrine of Virtue, Kant uses most of the time the End-in-Itself Formula to justify various ethical duties. I do not think this means Kant is on his way towards an ethics without the Universal Law formula; instead, I think that, for the purpose of justifying ethical duties in that context, he finds the End-in-Itself formula more appropriate.

What I have tried to show in this paper is that a question mark still remains over the extent to which Kant abandons the Formula of Autonomy. As mentioned in the introduction to this paper, my claim is not that I would have demonstrated Kleingeld was wrong to assert that Kant found the Formula of Autonomy obsolete and abandoned it; my claim was that, even if Kant did think he was no longer able to sustain this formulation of the Categorical Imperative and decided to abandon it, he did not need to.