Kant on Autonomy of the Will

Draft of a chapter to be included the Routledge Handbook of Autonomy (ed. Ben Colburn)

Biographical Note: Janis David Schaab is a Postdoctoral Researcher in Philosophy at the

University of Groningen. He received his PhD from the Universities of St Andrews and Stirling

in 2019. His research primarily focuses on Kantian approaches to ethical theory.

Abstract: Kant takes the idea of autonomy of the will to be his distinctive contribution to moral

philosophy. However, this idea is more nuanced and complicated than one might think. In this

chapter, I sketch the rough outlines of Kant’s idea of autonomy of the will while also

highlighting contentious exegetical issues that give rise to various possible interpretations. I

tentatively defend four basic claims. First, autonomy primarily features in Kant’s account of

moral agency, as the condition of the possibility of moral obligation. Second, autonomy

amounts to a metaphysical property as well as a normative principle and a psychological

capacity. Third, although there is legitimate scholarly disagreement about whether or not

autonomy involves self-legislation of the moral law, there is good reason to believe it

underwrites an ‘inside-out’ (as opposed to ‘outside-in’) conception of the relationship

between the will and moral requirements. Fourth, persons have dignity because their

autonomy makes them members in the set of beings over whom the categorical imperative

requires us to universalise our maxims, not because autonomy is an independently important

property.

1
1. Introduction

In the Groundwork of the Metaphysics of Morals, Kant boldly asserts that all previous attempts

“to detect the principle of morality […] had to fail” for the same reason (G 4:432) 1:

One saw the human being bound to laws by his duty, but it did not occur to anyone

that he is subject only to his own and yet universal legislation, and that he is only

obligated to act in conformity with his own will which is, however, universally

legislating according to its natural end.

According to Kant, therefore, the moral philosophers who came before him failed because

they did not consider the possibility that the human being possesses autonomy of the will,

which he defines as “the characteristic of the will by which it is a law to itself (independently

of any characteristic of the objects of willing)”, and that the principle of morality is a “principle

of autonomy” (G 4:440).

Since Kant thus takes the idea of autonomy of the will to be his distinctive contribution to

moral philosophy, we might expect this idea, and the role it plays in his moral theory, to be

clear and straightforward. This expectation might be reinforced by our confidence that we

have a decent intuitive grasp of the concept of autonomy—which, after all, is central to much

of current moral and political philosophy. However, Kant’s idea of autonomy of the will is more

nuanced and complicated than one might think.

1
I use the following abbreviations for (and translations of) Kant’s works: Col for Collins’s Lecture Notes (Kant

1997); CPR for Critique of Pure Reason (Kant 1999); CPrR for Critique of Practical Reason (Kant 1996); G for

Groundwork of the Metaphysics of Morals (Kant 2011); MM for Metaphysics of Morals (Kant 1996); Vigil for

Vigiliantius’s Lecture notes (Kant 1997). I follow the pagination of the Prussian Academy Edition, except for the

Critique of Pure Reason, where I follow the traditional A/B pagination.

2
In this chapter, I sketch the rough outlines of Kant’s idea of autonomy of the will while also

highlighting contentious exegetical issues that give rise to various possible interpretations. I

will start by considering why Kant attributes such central importance to autonomy of the will

(Section 2). On this basis, I try to gain a better understanding of what Kant means by

‘autonomy of the will’, particularly whether this term denotes a metaphysical property, a

normative principle, a psychological capacity, or some combination of these (Section 3).

Subsequently, I discuss whether and in what sense Kant’s doctrine of autonomy should be

understood to entail that we self-legislate the moral law (Section 4). Finally, I explore the

relationship between autonomy and the moral standing of persons as ends in themselves

(Section 5).

My focus throughout this chapter is on Kant’s conception of autonomy of the will. I thus

bracket Kant’s remarks about the autonomy of states. Moreover, my goal is to sketch what

autonomy of the will means for Kant and what role it plays in his foundational moral theory,

as laid out, primarily, in the Groundwork of the Metaphysics of Morals and the Critique of

Practical Reason. I do not discuss how Kant argues for the claim that we possess autonomy of

the will, let alone how this claim relates to his overall metaphysical views. Finally, I do not

consider how autonomy features in Kant’s works on aesthetics, anthropology, or political

philosophy. Any serious investigation of these issues would soon reveal that Kant’s idea of

autonomy is even more complex than I am able to show here.

2. Why Autonomy?

Why, according to Kant, does the fact that previous moral philosophers did not consider

autonomy of the will as a possibility mean that their attempts to determine the principle of

morality had to fail? The answer is that Kant holds that a “mere analysis of the concepts of

2
morality” (G 4:440) reveals that we must possess autonomy of the will if morality is to be more

than a “chimerical idea without truth” (G 4:445). That is, the idea of autonomy is already

entailed by ordinary people’s understanding of morality. In particular, autonomy is necessary

to make sense of the idea of moral obligation, which is central to this understanding.

It is sometimes stated that the reason why Kant deems autonomy necessary for moral

obligation is that he regards obligations as unconditionally binding. However, while this is

basically correct, it is oversimplifying and potentially misleading. Without further clarification,

the claim that moral obligations are unconditionally binding could be understood as the mere

claim that they bind us regardless of our antecedent desires. But intuitionists like W.D. Ross,

for example, agree with this claim and nevertheless deny that our will is a law to itself (Ross

2002: 157-160; see also Stratton-Lake 2013: 254-55). Ross holds that some acts are required

or prohibited as a matter of “the fundamental nature of the universe” (ibid: 29). Thus, to

understand why Kant attaches such great importance to autonomy, we need to determine in

what sense he takes moral obligations to be unconditionally binding. And to determine this,

we need to look at the entire bundle of moral concepts that he analyses in the Groundwork.

Kant’s analysis starts with the concept of the good will. He asserts that “[i]t is impossible to

think of anything at all in the world, or indeed even beyond it, that could be taken to be good

without limitation, except a GOOD WILL” (G 4:393). Kant grants that there are many other

things that can be taken to be good, such as wit, confidence, and happiness. Yet, Kant argues,

none of these things are good unconditionally. More specifically, they are not good unless they

occur in the presence of a good will.

According to Kant, to unpack the concept of a good will, we should “inspect the concept of

DUTY, which contains that of a good will, though under certain subjective limitations and

3
hindrances” (G 4:397). To imperfect rational beings like us, who are not solely determined by

reason but also possess a sensuous nature, the actions that a good will would perform are

represented as duties. More specifically, since our sensuous nature makes us susceptible to

the influence of inclinations, the concept of a good will features as a constraint or

“necessitation” in our deliberation (G 4:413). In short, it obligates us.

Kant argues that the limitations of imperfectly rational beings implicit in the concept of duty

allow us to bring the concept of a good will into especially sharp relief (G 4:397). Imperfectly

rational beings can act either from duty or from inclination. Thus, by contrasting the two

sources of motivation available to them, we can get a better grasp of what it would mean to

have a good will.

Kant argues that action “from duty” is to be distinguished, not only from action “contrary to

duty”, but also from action that is merely “in conformity with duty” (G 4:397-98). One of the

examples that Kant uses to illustrate the latter distinction is that of an agent who helps others

from sympathy, i.e. because their suffering pains him and its relief gratifies him. Kant asserts

that, while this agent’s action “conforms with duty”, is even “amiable”, it “has no true moral

worth” (G 4:397). While the agent acts as duty requires, he does so only because, and insofar

as, duty aligns with his inclinations. By contrast, the beneficent action of an agent who,

through grief or by nature, is not inclined to act beneficently at all, has “far higher worth”

(ibid.).

Kant takes this to show that, in imperfectly rational beings like us, a good will corresponds to

the will of an agent who does his duty from duty alone, and thus regardless of inclination. He

also takes it to show that “an action from duty has its moral worth not in the purpose that is

4
to be attained by it, but in the maxim according to which it is resolved upon, and thus […] the

principle of willing according to which […] the action is done” (G 4:399-400).

Kant’s reasoning seems to run as follows. The difference in moral worth between an action

from duty and an action from inclination cannot be due to their respective purposes because

they can have the same purpose (see Korsgaard 2008: 178; Timmermann 2007: 38). For

example, both the sympathetic benefactor and the agent who acts beneficently from duty

intend to relieve suffering.

Connectedly, for Kant, an action’s purpose is problematically external to the agent’s

deliberation. In particular, an action’s purpose or, more generally, an “object of the

desiderative faculty” cannot determine the will immediately (G 4:400). If it determines the will

at all, it does so via an inclination of the agent’s. And we have already seen that, for Kant,

actions are done either from inclination or from duty.

Note that the properties of the particular act-type that the agent performs are external to his

deliberation in the same way as his purpose. Thus, they cannot be what determines the choice

of a good-willed agent either. Accordingly, Kant remarks that moral worth does not reside in

“the matter of the action” (G 4:416).

For Kant, then, actions from duty are guided, not by some external object or feature, but by

something internal to the agent’s deliberation. More specifically, they are guided by the

quality of the agent’s maxim. He states that “[a] maxim is the subjective principle of willing”

(G 4:400, footnote). He later adds that a maxim is the principle an agent actually acts on, and

thus to be distinguished from the law, which is the principle that the agent ought to conform

to (G 4:420-21, footnote). What exactly Kant takes maxims to be is a difficult exegetical

question (see Gressis 2010a; Gressis 2010b). What is important for our purposes is that Kant

5
thinks that the moral worth of an action depends on which maxim governs the agent’s choice

of the action.

But what characterises the maxim underlying a morally good action? Kant writes that “as every

material principle has been taken away from it […] nothing remains for the will that could

determine it except, objectively, the law and, subjectively, pure respect for this practical law”

(G 4:400). Kant’s reasoning is that, since a morally worthy action is done from duty, and duty

presupposes a law,2 the maxim underlying such an action must be lawful. Therefore, an action

from duty is an action from respect for the law.

However, the law in question cannot be a particular, substantive law. This is what Kant’s

statement that “every material principle has been taken away from [the will]” highlights

(ibid.). It means that, in choosing how to act, a good will is not guided by the satisfaction of

immediate inclination, the properties of a particular act-type, or the realisation of some

purpose. Instead, the maxims underlying morally worthy actions are distinguished by their

“mere conformity with law as such”, i.e. their conformity to the very idea of law (G 4:402). In

other words, what distinguishes these maxims is that they could serve or function as laws. We

might say that they possess the form of laws (see G 4:436).

According to Kant, then, the lawful form of its maxims “alone is to serve the will as its principle,

i.e. I ought never to proceed except in such a way that I could also will that my maxim should

become a universal law” (G 4:402).3 Kant thus identifies the merely formal requirement only

2
“The concept of duty stands in immediate relation to a law (even if I abstract from all ends, as the matter of the

law)” (MM 6:389).

3
Why does Kant suddenly speak of universal law here? The reason seems to be that the law in question must be

valid for all rational beings, independently of inclinations (see G 4:389).

6
to act on maxims that one could also will to become universal laws as the supreme principle

of morality. For imperfectly rational beings, who are necessitated by the law, this principle

takes the form of an imperative. Kant later identifies it as the only imperative that necessitates

us independently of any inclinations and is therefore called “categorical” rather than

“hypothetical” (G 4:420-21). He states the categorical imperative as follows: “act only

according to that maxim through which you can at the same time will that it become a

universal law” (G 4:421).

The categorical imperative leads us straight to the idea of autonomy of the will. Recall that

Kant defines autonomy as “the characteristic of the will by which it is a law to itself

(independently of any characteristic of the objects of willing)” (G 4:440). This is exactly the

characteristic that the will must have if it is to be able to act from duty: it must select its

maxims by a law, and yet independently of any particular, substantive law. To do so, it must

select its maxims in light of whether they could be willed to be universal laws. Accordingly,

Kant also refers to the categorical imperative as “the principle of every human will as a will

universally legislating through all its maxims” (G 4:432) or simply as “the principle of

autonomy” (G 4:440).

We can now see why it is so important to Kant that the will is regarded as autonomous.

According to Kant, the concepts of duty and the good will, which are central to ordinary

people’s understanding of morality, entail the concept of a law that binds unconditionally.

Such a law is not merely independent of our antecedent desires (pace Ross), but necessitates

us through pure respect for lawfulness as such. But this means that, if the concepts of duty

and a good will are instantiated in the real world, we must possess autonomy of the will.

Therefore, Kant’s charge that previous thinkers foreclosed the possibility of autonomy

7
amounts to the charge that they rendered morality itself, as understood by ordinary people,

a mere fiction.

3. The Nature of Autonomy

Kant’s idea of autonomy primarily characterises the way we, as agents, must relate to moral

requirements if such requirements are to be possible. This, then, is the primary function that

autonomy serves within Kant’s framework. But what exactly does autonomy of the will

amount to? Is it a metaphysical property, a normative principle, a psychological capacity, or

some combination of these?

Kant argues that autonomy is both necessary and sufficient for freedom of the will. The will is

free if it “can be efficient independently of alien causes determining it” (G 4:446). Thus, if the

will can determine itself by its own lawful form, and thus independently of sensibility, it is free.

Therefore, autonomy is sufficient for freedom of the will.

To show that autonomy is also necessary for freedom of the will, Kant argues that freedom

has not only a “negative” but also a “positive” aspect (ibid.). According to Kant, “the will is a

kind of causality” (ibid.). Now, “[s]ince the concept of causality carries with it that of laws”

connecting causes with effects, this means that a free will cannot be “lawless” (ibid.).

Accordingly, a free will is not merely independent of alien causes (and thus of the laws of

nature) but proceeds in accordance with its own law. That is, it is autonomous.

As Karl Ameriks points out, Kant’s notion of freedom of the will is “transcendental” rather than

“empirical” (Ameriks 2013: 54-58). For the will to be independent from ‘alien causes’—and

thus to exhibit the negative aspect of freedom—it is not sufficient that it be independent from

empirical causes outside ourselves, such as manipulation or external compulsion. Instead, the

8
will must not be determined by any empirical causes, including aspects of our empirical nature

that we might ordinarily regard as expressing our individual nature or identity.

Thus, the association between autonomy and freedom of the will clearly shows that autonomy

has a metaphysical dimension. But the association between autonomy and transcendental

freedom also highlights another, normative dimension of autonomy (see, e.g., Reath 2013: 36;

Sensen 2013b: 263). We can see this by shifting our attention to the side of autonomy that

mirrors the positive aspect of freedom. Since autonomy is the characteristic of the will to be

a law to itself, independently of the objects of volition, autonomy is not only necessary but

sufficient for the authority of the moral law. Thus, Kant writes: “a free will and a will under

moral laws are one and the same” (G 4:447). Henry Allison has labelled this the “reciprocity

thesis” (1990: Chapter 11).

It follows that imperfect rational beings like us (who do not necessarily conform to the moral

law), insofar as they are autonomous, are bound by the law, in the form of the categorical

imperative. Thus, our will’s being a law to itself—its ability to be efficacious independently of

empirical causes—has implications for what we ought to do. As Jens Timmermann puts it,

“[o]ddly enough, ‘can’ implies ‘ought’” (2007: 121). We might even say that, for beings like us,

autonomy itself is a normative principle. Indeed, Kant explicitly identifies “[t]he autonomy of

the will as the supreme principle of morality” (G 4:440). For Kant, to say that the will is

autonomous just is to say that it is subject to the “principle of autonomy”, i.e., “not to choose

in any other way than that the maxims of one’s choice are also comprised as universal law in

the same willing” (ibid.).

This highlights that autonomy is not only the condition of the possibility of moral obligation

but also furnishes the content of moral commands. Since the moral law’s only prescription is

9
that our maxims be willable as universal laws, we might say that the moral law commands

that we “realise” or “express” our autonomy (Sensen 2011: 169 and Hill 1992: 85,

respectively). In Kant’s words, the law “commands neither more nor less than just this

autonomy” (G 4:440).

But the positive aspect of freedom not only reveals the normative dimension of autonomy; it

also implies that beings with autonomy must have certain psychological capacities. In

particular, if a rational will is not only (negatively) independent of the laws of empirical nature

but also (positively) a law to itself, rational agents must be capable of recognising and being

motivated by that law. It thus seems that autonomy is not only a metaphysical property and a

normative principle, but also “a recognitional cum motivational capacity” (Reath 2013: 33).

Note, however, that the precise psychological implications of autonomy can vary between

different kinds of rational beings. For imperfect rational beings like us, who are sensuously

affected, the capacity to be motivated by the law requires the existence of an incentive to

comply with it (CPrR 5:72). Since this incentive must be able to counteract impulse and

inclination, which are “based on feeling”, the incentive must itself be (based on) a feeling

(ibid.). Yet, since this feeling could not enable autonomy if it was empirically conditioned, it

must instead be “solely produced by reason” (CPrR 5:76). Kant identifies that feeling as

“respect for the moral law” (ibid.).

By contrast, perfect rational beings (such as “the divine will”) do not require any incentive to

be motivated by the moral law (CPrR 5:72). Since they are not sensuously affected, their

actions necessarily conform to the moral law. Consequently, autonomy should not be

identified with the capacity to act from respect for the law, although it requires that capacity

in human beings.

10
4. Autonomy as Self-Legislation of the Moral Law?

So far, my focus has been on the autonomous will’s ability to be determined by a law that is

not empirically conditioned. But ‘autonomy’ literally means self-legislation. As Kant himself

states, “the will is not just subject to the law, but subject in such a way that it must also be

viewed as self-legislating, and just on account of this as subject to the law” (G 4:431). Hence,

autonomy of the will seems to imply that the will in some sense gives law to itself.

This dimension of autonomy seems important because, as Andrews Reath observes, it

suggests that autonomy underwrites “a normative standing of the rational will” (2013: 33).

The autonomous will seems to be “normatively self-governing” in that “it need not answer to

any outside authority, but only to itself” (ibid.) In short, autonomy underwrites a kind of

“sovereignty of the will over itself” (Reath 2013: 49).

However, the idea that the will self-legislates the moral law, i.e. the law expressed by the

categorical imperative, can seem puzzling. The puzzle is how to square this idea with the

unconditional authority of the moral law. For talk of ‘self-legislation’ may seem to suggest that

the law’s authority is conditional on an act of legislation (Kleingeld and Willaschek 2019: 3;

Pinkard 2002: 59; Stern 2012: 13-14).

This raises the question of how the ‘legislation’ part of ‘self-legislation’ is to be understood.

Kant distinguishes two roles in legislative processes. He calls the one who commands a law its

“lawgiver” (MM 6:227). Since the lawgiver makes the law binding, he is the “author of the

obligation” (ibid.). But the lawgiver need not also be the “author of the law”, i.e. the one who

determines its content (ibid.) And, indeed, Kant seems to suggest that the moral law has no

author (Col 27: 282–83; Vigil 27:544; see also, e.g., Kain 2004; Stern 2010; Timmermann 2007:

11
107; Wood 2008: 113). It is thus not up to us, nor anyone else, what the law commands.

Instead, the content of the law is given by reason alone.

However, this qualification may not dispel the air of paradox surrounding the notion of

autonomy completely. After all, it leaves open the possibility that it is up to us—the very

agents who are subject to the law—to make the law binding. This leads us to a second

question: who or what does the ‘self’ in ‘self-legislation’ refer to?

One option is that it refers to our self as concrete, empirical human beings. This interpretation

might be congenial to theorists who associate autonomy with the expression of one’s personal

values or identity (e.g. Bratman 1979; Frankfurt 1988). And, indeed, some contemporary

Kantians are sometimes read as endorsing it (e.g. Korsgaard 2009). But this interpretation also

makes it harder to see how the moral law can possess unconditional authority. For it suggests

that the self that is subject to the moral law is indeed identical to the self that legislates it.

Kant seems to rule out this interpretation for this very reason (MM 6:417).

It seems more likely that, instead, the ‘self’ refers to our noumenal self (Sensen 2013b: 268;

Timmerman 2007: 115). While our phenomenal self corresponds to our nature as a concrete

human being, our noumenal self corresponds to our nature as a being with pure practical

reason, an immutable nature that we share with all other rational beings (MM 6:418). A

related (though not identical) distinction in Kant’s works is between our faculty of “choice”

(Willkür) and our “will” (Wille): while the former enables us to act and realise the objects of

our desires in general, the latter enables us to be guided by pure reason alone (MM 6:213;

also see Sensen 2013b: 269; Timmermann 2007: 115, 173-75). Thus, if our noumenal self or

will—as opposed to our phenomenal self or faculty of choice—gives the law to us, this might

12
explain how we can be both self-legislating and subject to the unconditional authority of the

law.4

A third option is that the ‘self’ in ‘self-legislation’ does not refer to one of the agent’s selves

at all. Instead, it indicates that the law is authoritative of itself, i.e. without presupposing

anything else (Sensen 2013b: 269; also see O’Neill 2003: 15-19). The idea is that the moral law

states the conditions of the possibility of lawgiving, and therefore does not depend on the

authority of any legislator.

This interpretation suggests that the moral law is not ‘legislated’ in any literal or

straightforward sense. But it does not abandon the assumption that the moral law is legislated

in some sense, which is shared by a wide range of scholars (e.g. Engstrom 2009: 149; Reath

2006; Reath 2013; Schneewind 1998: 6; Wood 2008: 106). An outright rejection of this

assumption has been suggested more recently, by Pauline Kleingeld and Marcus Willaschek

(2019).

Kleingeld and Willaschek raise a third question about talk of ‘self-legislation’: what, if anything,

is being legislated? They argue that Kant does not regard the moral law, as expressed by the

categorical imperative, as self-legislated at all (2019: 2). He unequivocally speaks of self-

legislation only in the context of substantive moral laws, i.e. imperatives that express the

specific duties that result from applying the categorical imperative to our maxims. According

to Kleingeld and Willaschek, Kant calls the categorical imperative “the principle of autonomy”

because it commands that we select our maxims as if we were thereby also giving law to all

rational beings (2019: 6-7). They conclude that Kant’s notion of autonomy does not imply that

4
For a qualified criticism of this idea, see Reath (2013: 45-47).

13
the authority of the moral law is grounded in any kind of legislation. Instead, this authority is

a fundamental, a priori fact of reason (2019: 13-16).

The latter three interpretations of the notion of self-legislation—that the noumenal self or

Wille legislates the law, that the law is authoritative of itself, and that only particular moral

laws are to be regarded as self-legislated—all seem like promising candidates when it comes

to salvaging the unconditional authority of the moral law. However, these interpretations may

not seem congenial to the idea with which this section started: that autonomy underwrites a

kind of sovereignty of the will over itself.

Yet, one idea, which is popular among Kantian ethicists and seems largely compatible with

these three interpretations, might help to alleviate this concern. The idea is that the moral law

is the constitutive principle of willing or, equivalently, of the activity of practical reason.

Different versions of this idea have been proposed in the literature (see, e.g., Engstrom 2009;

Hill 1992: 87-88; Korsgaard 2009; Reath 2013; Sensen 2019).5 Here, I merely try to sketch what

I take to be their unifying theme.

As Christine Korsgaard notes, to say that a principle is constitutive of an activity is to say that

“unless you are guided by the principle in question, you are not performing that activity at all”

(2009: 29). For example, the constitutive principle of building a house requires that you erect

a physical structure that provides shelter from the weather (ibid.). If you do not at least aspire

or strive to conform to this principle, you are not engaged in house-building at all.

5
One question that divides theorists is whether the law is constitutive of Wille (Sensen 2019), Wille and Willkür

combined (Reath 2013), or all types of practical reason (Korsgaard 2009).

14
The constitutive principle of an activity is a normative standard for that activity (Korsgaard

2009: 29-32). For example, building a physical structure that has random holes in the walls

and ceiling will at best count as building a house badly because it at most imperfectly conforms

to the constitutive principle of that activity. However, whereas normative standards that are

not constitutive of house-building (e.g. standards of beauty) depend on the presence of

certain external factors (e.g. whether the house-builder cares about beauty), the constitutive

principle of an activity seems to provide a normative standard which is internal to that activity.

Thus, the claim that the moral law is the constitutive principle of willing amounts to the claim

that, unless we acknowledge the categorical imperative as binding, we are not engaged in the

activity of willing at all. What is more, it implies that the categorical imperative provides a

normative standard for our actions which is internal to our will. Far from imposing an external

constraint on us, it expresses our own will’s in-built ambition to be a law to itself.6

It is not implausible to attribute this view to Kant. As we saw above, Kant thinks that if our will

possesses the metaphysical property of transcendental freedom (and the corresponding

psychological capacity), we are bound by the normative requirement to make our maxims

willable as universal laws. The idea that the categorical imperative is constitutive of the activity

of a free will can help us make sense of this.

If this is Kant’s view, then perhaps the idea that autonomy accords the will a kind of

sovereignty can be salvaged after all. Even if we do not legislate the moral law to ourselves, it

is in a sense our own law. We strive to comply with this law by virtue of being the kind of

6
This need not make immoral willing impossible. Agents who don’t comply with the moral law might still be said

to acknowledge its bindingness. Kant writes: “even the most hardened scoundrel [...] is conscious of [...] the law,

the repute of which he recognizes as he transgresses it” (G 4:454-55).

15
creatures who engage in willing (see Reath 2013). Indeed, since it is an internal standard of

willing, the law would not bind anyone if there were no creatures like us (see Sensen 2013c;

Sensen 2019).

This suggests that Kant’s idea of autonomy of the will parallels his so-called Copernican

Revolution in theoretical philosophy (see CPR BXvi). As Oliver Sensen puts it, just as Kant’s

theoretical philosophy proposes that “we are not merely passive observers in cognizing

nature, but that our mind (partly) constitutes how reality appears to us”, his practical

philosophy proposes that “it is an activity of reason, something reason does spontaneously

and out of itself (not something that it discovers) that grounds morality” (2019: 199 and 217,

respectively). We might say that Kant’s idea of autonomy portrays the relationship between

the will and moral requirements as ‘inside-out’ rather than ‘outside-in’.

5. Autonomy and Dignity

I have argued that autonomy primarily features in Kant’s theory of moral agency. But Kant

famously claims that rational beings are “persons” rather than “things”, “because their nature

already marks them out as ends in themselves, i.e. as something that may not be used merely

as a means” (G 4:428). And he writes that “[a]utonomy is […] the ground of the dignity of a

human and of every rational nature” (G 4:436). This suggests that autonomy also plays an

important role in Kant’s account of moral patienthood.

Furthermore, since Kant describes “dignity” as “an inner worth”, contrasting it with “a price”,

which is merely “a relative worth” (G 4:435), it might seem as if the property of autonomy has

value or some other kind of moral importance, independently of the moral law. And this might

in turn seem to explain the special moral standing of persons. Some Kant scholars seem to

come close to endorsing this interpretation (e.g. Caranti 2017; Guyer 1998). It might appeal

16
to theorists who want to derive human dignity or human rights from autonomy directly,

without taking a detour through the categorical imperative (e.g. Griffin 2008). But this

interpretation seems to contradict my claim that Kant proposes an ‘inside-out’ rather than an

‘outside-in’ relationship between rational agents and moral requirements.

However, there is good reason to doubt that Kant conceives of autonomy as an independently

important property. As we saw in Section 2, Kant thinks that a morally worthy action cannot

be one that is done for the sake of some “object of the desiderative faculty”, i.e. some object

that lies outside the will and that the will is aimed at, because such an object can only

determine the will via the agent’s inclinations (G 4:400; also see CPrR 5:62-64). Connectedly,

actions done for the sake of such an object would not realise our autonomy. After all, when

we act from inclination, we conform to the laws of empirical nature.

These considerations also apply to the notion of autonomy as an independently important

property (Sensen 2013b: 276-79).7 Any action for the sake of such a property would be from

inclination, and hence not realise autonomy. Kant holds that autonomy, and the ordinary

conception of morality that relies on it, are possible only if the will has its own, a priori law.

All objects of the will—even the property of autonomy itself—are to be assessed by that law.

But how is the moral importance of autonomy to be understood, if not in terms of an

independently important property? Recall that the moral law requires that we act on maxims

that we could will to be universal laws, i.e. laws that determine the will of every rational being.

Hence, our will “must always take its maxims from the point of view of itself, but also at the

same time of every other rational being as legislating (which are therefore also called

7
Also see Chapter 3 in this volume.

17
persons)” (G 4:438). In other words, the will’s maxims must be such that they could be willed

as universal laws by—and thus realise the autonomy of—every rational being. And this,

arguably, just is the requirement not to use others as mere means and instead to regard them

as ends-in-themselves. The following passage seems to confirm this reading (CPrR 5:87, italics

altered):

[E]very will […] is restricted to the condition of agreement with the autonomy of the

rational being, that is to say, such a being is not to be subjected to any purpose that is

not possible in accordance with a law that could arise from the will of the affected

subject himself; hence this subject is to be used never merely as a means but as at the

same time an end.

The idea seems to be roughly that, in order to avoid using someone as a mere means, we have

to act on a maxim that they could also rationally will (O’Neill 1989: 141-44; Reath 2013: 51-

52; Sensen 2011: 109).8 And this is just what it takes for our maxim to be compatible with the

categorical imperative.

Therefore, persons have a special importance or “worth”9, not because autonomy is an

independently valuable or otherwise important property, but because the law commands that

our maxims be such that we could will them to become universal law, which is the case if and

only if they could be willed by any rational being as such. The dignity of persons is therefore

grounded in their autonomy in the sense that their autonomy makes them members in the

set of beings over whom our maxims must be universalisable (see Bader forthcoming). In

8
For a different reading, see Kleingeld (2020). For discussion, see Chapter 3 in this volume.

9
I bracket the question of whether the moral standing of persons should be framed in terms of value at all (see

Bader forthcoming).

18
Kant’s words, autonomy makes each person a “legislating member of the universal kingdom

of ends” (G 4:438).

This corroborates my claim that autonomy primarily characterises the way persons,

considered as moral agents, relate to moral requirements—namely as (fellow) legislators

rather than as mere subjects. The moral standing of persons, considered as moral patients,

merely reflects this relation. Hence, Kant writes, “morality and humanity, in so far as it is

capable of morality, is that which alone has dignity” (G 4:435, italics added).

6. Conclusion

I have argued that autonomy primarily features in Kant’s account of moral agency, as the

condition of the possibility of moral obligation. Autonomy is at once a metaphysical property

of the will, a normative principle (for imperfectly rational beings) and a psychological capacity.

While there is reasonable disagreement about whether autonomy implies that the moral law

is self-legislated, there is good reason to think that it underwrites an ‘inside-out’ picture of the

relation between moral agents and moral requirements. These claims are not undermined by

autonomy’s relation to dignity. Persons have dignity because their autonomy makes them

members in the set of beings over whom the law requires us to universalise, not because

autonomy is an independently important property.

Further Reading

Oliver Sensen (ed.), Kant on Moral Autonomy (Cambridge: Cambridge University Press, 2013)

is a useful collection of recent essays on Kant’s conception of autonomy, its historical

significance, and its relevance today. Stefano Bacin and Oliver Sensen (eds.), The Emergence

of Autonomy in Kant’s Moral Philosophy (Cambridge: Cambridge University Press, 2019) is a

19
collection of cutting-edge essays on the development of the idea of autonomy across the

various stages of Kant’s ethical thought.

References

Allison, Henry. 1990. Kant’s Theory of Freedom, Cambridge: Cambridge University Press.

Ameriks, Karl. 2013. “Vindicating Autonomy,” in Sensen 2013a: 53-70.

Bader, Ralf. Forthcoming. “The Dignity of Humanity,” in Sarah Buss and Nandi Theunissen

(eds.), Rethinking the Value of Humanity, Oxford: Oxford University Press.

Bratman, Michael. 1979. “Practical Reasoning and Weakness of the Will.” Noûs, 13: 131–51.

Caranti, Luigi. 2017. Kant’s Political Legacy: Human Rights, Peace and Progress, Cardiff:

University of Wales Press.

Engstrom, Stephen. 2009. The Form of Practical Knowledge, Cambridge, MA: Harvard

University Press.

Frankfurt, Harry. 1988. The Importance of What We Care About, Cambridge: Cambridge

University Press.

Gressis, Robert. 2010a. “Recent Work on Kantian Maxims I: Established Approaches.”

Philosophy Compass 5(3): 216–27.

— 2010b. “Recent Work on Kantian Maxims II.” Philosophy Compass 5(3): 228–239.

Griffin, James. 2008. On Human Rights, Oxford: Oxford University Press.

Guyer, Paul. 1998. “The Value of Reason and the Value of Freedom.” Ethics 109: 22–35.

Hill, Thomas Jr. 1992. “The Kantian Conception of Autonomy,” in Dignity and Practical Reason

in Kant’s Moral Theory, Ithaca and London: Cornell University Press: 76-96.

Kain, Patrick. 2004. “Self-Legislation in Kant’s Moral Philosophy.” Archiv für Geschichte der

Philosophie 86: 267-306.

Kant, Immanuel. 1996. Practical Philosophy, Mary Gregor (ed. and trans.), Cambridge:
20
 Cambridge University Press.

— 1997. Lectures on Ethics, Peter Heath (ed. and trans.) and Jerome B. Schneewind (ed.),

Cambridge: Cambridge University Press.

— 1999. Critique of Pure Reason, Paul Guyer and Allen W. Wood (eds. and trans.),

Cambridge, UK: Cambridge University Press.

— 2011. Groundwork of the Metaphysics of Morals: A German-English Edition, Mary

Gregor and Jens Timmermann (eds. and trans.), Cambridge: Cambridge University

Press.

Kleingeld, Pauline. 2020. “How to Use Someone ‘Merely as a Means’.” Kantian Review 25(3):

389-414.

Kleingeld, Pauline and Willaschek, Marcus. 2019. “Autonomy Without Paradox: Kant, Self-

Legislation and the Moral Law.” Philosophers’ Imprint 19(6): 1-18.

Korsgaard, Christine. 2008. The Constitution of Agency: Essays on Practical Reason and Moral

Psychology, Oxford: Oxford University Press.

— 2009. Self-Constitution: Agency, Identity, and Integrity, Oxford: Oxford University

Press.

O’Neill, Onora. 1989. Constructions of Reason: Explorations of Kant’s Practical Philosophy,

Cambridge, UK: Cambridge University Press.

— 2003. “Autonomy: The Emperor’s New Clothes.” Proceedings of the Aristotelian

Society, Supplement, 77: 1-21.

Pinkard, Terry. 2002. German Philosophy 1760–1860: The Legacy of Idealism, Cambridge:

Cambridge University Press.

Reath, Andrews. 2006. Agency & Autonomy in Kant’s Moral Theory, Oxford: Oxford University

Press.

21
 — 2013. “Kant’s Conception of Autonomy of the Will,” in Sensen 2013a: 32-52.

Ross, William D. 2002. The Right and the Good, Philip Stratton-Lake (ed.), Oxford: Clarendon

Press.

Schneewind, Jerome B. 1998. The Invention of Autonomy: A History of Modern Moral

Philosophy, Cambridge: Cambridge University Press.

Sensen, Oliver. 2011. Kant on Human Dignity, Berlin/Boston: De Gruyter.

— 2013a (ed.). Kant on Moral Autonomy, Cambridge: Cambridge University Press.

— 2013b. “The Moral Importance of Autonomy,” in Sensen 2013a: 262-81.

— 2013c. “Kant’s Constructivism”, in Carla Bagnoli (ed.), Constructivism in Ethics,

Cambridge, UK: Cambridge University Press: 63-81.

— 2019. “Kant’s Constitutivism,” in Robinson dos Santos and Elke Elisabeth Schmidt

(eds.), Realism and Antirealism in Kant’s Moral Philosophy: New Essays, Berlin/Boson:

De Gruyter, 197-220.

Stern, Robert. 2010. “Moral Scepticism and Agency: Kant and Korsgaard.” Ratio 23(4): 453-

474.

— 2012. Understanding Moral Obligation: Kant, Hegel, Kierkegaard, Cambridge:

Cambridge University Press.

Stratton-Lake, Philip. 2013. “Morality and Autonomy,” in Sensen 2013a: 246-61.

Timmermann, Jens. 2007. Kant’s Groundwork of the Metaphysics of Morals: A Commentary,

Cambridge, UK: Cambridge University Press.

Wood, Allen W. 2008. Kantian Ethics, Cambridge, UK: Cambridge University Press.

22