About the Groundwork

Section I

The Good Will
The End of Practical Reason
Duty and Moral Worth
The Three Propositions
The Feeling of Respect
Conformity to Universal Law
Maxims of Action
The Lying Promise
The Inquiring Murderer

Section II

Hypothetical Imperatives
The Categorical Imperative
The Suicide Example
Perfect and Imperfect Duties
The Contradictions
Moral Realism
The Formula of Humanity
Deontology and Ethical Ends
The Formula of Autonomy
Autonomy and Heteronomy
The Kingdom of Ends

Section III

Freedom of the Will
Free in a Practical Respect
The Hidden Circle
Two Standpoints
The Deduction
Failure of the Deduction


Freedom of the Will “Natural necessity was a heteronomy of efficient causes, since every effect was possible only in accordance with the law that something else determines the efficient cause to causality; what, then, can freedom of the will be other than autonomy, the will’s property of being a law to itself?”

Two kinds of causality are conceivable for Kant: free causality and unfree causality. The principal difference between them, according to the quotation above (Kant 1997a, 52/4:446-7), depends on the source of the law of the causality. If the source is other than the cause itself, then the causality is unfree and the causal agent is heteronomous. But agents who are the source of the law of their causality, or of its bindingness on them, are free and autonomous. Autonomy—subjecting oneself to law—is therefore freedom.

The structure of causality. As Kant conceived of causality, all causes operate through their own powers. David Hume had denied this. Or he had at least denied that we could ever have experience of causal power. He argued that nothing can be experienced in the causal relation but the cause and the effect (fire and smoke); and that our habitually associating the two gives us the impression of their necessary connection (see Kant 1997b, 44-45/5:50-51). But Kant held that conceiving of two things as connected by the cause-effect relationship is conceiving of the one as bringing about the other by exercising its power. That is what it means to act, after all. The cause acts; and a law of action is necessary for every cause. For otherwise the action that brings about the effect could be attributable to any cause whatever, or random. Even from the cause’s own point of view its action would seem random if there were no law-like connection between its exercising its power and its bringing about a particular effect. Hence, the law of the cause’s operation is an essential part of its relationship to its effect. It cannot be its effect, an effect it is responsible for, without the law for the use of its power. The difference between autonomy and heteronomy, then, is the difference between the cause’s giving itself the law, and its being given the law from some other source: perhaps nature, or God.

Causality, freedom and time. How is it possible to give oneself a law for one’s actions, one’s causality? Wouldn’t that amount to giving oneself a law one prefers, and so the law’s being given, in reality, by the preference, or by whatever caused one to prefer it? This is the kernel of the problem of the will’s freedom. My desires and circumstances seem to cause my actions. I nevertheless feel that I am free to step back, and decide not to let my desires cause me to act—or not to let certain desires move me. But my decision, when I step back, would itself be caused by a preference, and so by a desire over which I had no control. So my “feeling of freedom” seems to be an illusion. I can’t get out of the sphere of being determined to action by my desires, and I do not freely choose my desires. Kant was convinced that this is exactly right, so long as we are talking about human life in the course of time—or what he called the “sensible world,” or the “world of appearances,” or “phenomena.” But it is a crucially important part of his wider philosophy that time is a feature of human perceptual consciousness, and so that everything we think of as “happening” is conditioned by human thought or understanding. God, for example, would not see things in time, in the human way. Being eternal, there would be no before and after for him. There is a before and after only from the perspective of a particular point in time, and being eternal, God would not occupy a particular point in time: he would occupy all time, or better, no time at all. For this reason God would not see or conceive of things as we time-bound human being do. He would not see things as appearances in time; he would see or conceive of things as they are in themselves. We cannot know what things in themselves may be like, because we cannot think what anything would be like except as conditioned by time. This distinction between things as they are in themselves and as they appear in human experience now opens up the prospect of freedom of the will. The problem posed by the concept of freedom is that no matter how I act, in time, I inevitably act on the basis of a prior desire, which I have not chosen—of if I have chosen it, I have not chosen the desire that caused me to chose it. That is how things go if I think of myself as appearing in time. But if I think of myself as a thing in itself, outside of time, then no prior desire could determine how I act. So although in the former perspective I must see myself as acting always on the basis of prior desires, in the latter perspective I must consider myself free from determination by prior desires (Kant 1998, A538-41/B566-9). Kant referred to this freedom as “negative.” It is not yet freedom to, but only freedom from.

Positive freedom. To “act” is to cause something—the causal relation is the cause’s acting through its own power to bring about an effect (Kant 1998, A204/B250). In order to be conceived as acting, the cause must have a law for its action, as explained above. If the law is given to it, or imposed upon it from elsewhere (heteronomy), then the cause does not act freely. Only the cause imposing the law upon itself acts freely; and this cannot be done in time. For in time, imposing a law upon oneself would be determined by a prior, unchosen desire. To be outside of time, and so free from prior desire to impose a law upon oneself, would be only negative freedom. Positive freedom would mean that one is free to act on the law of one’s causality—a law that would be given by oneself, which implies autonomy. So we see why Kant asks: “what, then, can freedom of the will be other than autonomy, the will’s property of being a law to itself?”

Autonomy vs. heteronomy, yet again. It is important not to think of acting on a desire as heteronomy, and so unfree, and of acting on a self-given law as autonomy, and so free. Thinking of the two this way makes it conceivable that one can act heteronomously one time, and autonomously the next. It also raises a very serious problem for Kantian ethics, insofar as it implies that we are unfree in, and so not morally responsible for, our heteronomous actions. For this reason, the following clarifications should inform our use of these terms:

“Heteronomy” characterizes a cause whose law of operation is imposed upon it. All natural causes are therefore heteronomous. When human beings are considered as wholly natural beings they are heteronomous, because nature imposes the laws of their causality upon them. They have, by nature, a set of desires making up their character, and all of their actions follow the laws of their characters.

“Autonomy” characterizes a cause whose law of operation is self-imposed, and thus, one which is both free from acting on prior desire, and free to act on its own law. When human beings are considered autonomous, they are conceived as outside of time. They must therefore be conceived as “always” autonomous. They must not be conceived as heteronomous at one time, and as autonomous at another time.

The way Kant saw it, if in our concept of “human being” we include the idea of freedom of the will, then we must regard human beings as autonomous. But if in our concept of the human being we leave out the idea of freedom of the will, and consider the human being as acting wholly in time, always on the basis of some prior desire, then in this case the human being is heteronomous. Autonomy and heteronomy are thus “global” characterizations of human beings. It is not that they are autonomous one time and heteronomous another. Kant put it this way:

The sensible nature of rational beings in general is their existence under empirically conditioned laws and is thus, for reason, heteronomy. The supersensible nature of the same beings, on the other hand, is their existence in accordance with laws that are independent of any empirical conidition and thus belong to the autonomy of pure reason. (Kant 1997b, 38/5:43)

Otherwise stated, “autonomy” and “heteronomy” apply to human beings depending on how we conceive of them. When they are considered free and morally responsible for their actions, and so as existing outside of time, they are autonomous. When they are considered as always determined to action by natural laws, including psychological laws of desire, and so as existing exclusively in time, they are heteronomous. One does not go back and forth, acting autonomously sometimes, and acting heteronomously other times. The human being we consider to be free and morally responsible for her conduct is an autonomous being. The actions of this being, whether they are wrong or right, desire-based or motivated by pure respect for the moral law, are neither autonomous nor heteronomous. These terms apply to agents, and sometimes to moral principles. They do not apply to actions (see Hill 1992, 85; for an opposing view see Korsgaard 2009, 81).

Freedom and determinism. Kant believed in the “compatibility” of freedom and determinism in precisely this sense: that it is not a contradiction to recognize (1) that everything that happens has a prior cause, and (2) that human actions arise through freedom of the will. But Kant rejected what is today known as compatibilism, or sometimes, soft determinism. This is the belief (1) that everything that happens has a prior cause, but (3) human beings can be free and responsible in their actions when these are caused or directed by internal forces. Kant did not think this idea of freedom deserved the name, referring to it as a “subterfuge,” and as “quibbling with words.” He said that if we accept this idea then we might as well attribute free will to a clock, since the movements of its hands are controlled by its own internal mechanism (Kant 1997b, 81/5:96). Yet statements (1) and (2) do seem contradictory; so how does Kant see them as compatible? The answer is that he restricts statement (1) to a description of the sensible world of appearances in time; but statement (2) is about another world. The freedom of human beings belongs to the “intelligible world” of thing in themselves, where there is no time; and—this is important—the actions of human beings in the sensible world are appearances of what they do in the intelligible world. Suppose I freely do action A in the intelligible world. Then, my A-ing appears “spread out,” as it were, through the span of time I occupy in the sensible world. All of my sensible actions are, in other words, so many appearances of the action that I do freely, outside of time, in the intelligible world. It is as if in that world my action is writing a script that I follow here in this world, like an actor on stage (McCarty 2009, 152-4). My writing the script could also be thought of as my formation of my character in the sensible world; and thus I can be held morally responsible for my actions in this world as determined by that character. If I am sometimes dishonest in the sensible world, it is because I, in the intelligible world, formed a character for myself that is less than perfectly honest.

Freedom and autonomy. Free will explained as timeless action establishing one’s character for conduct in the sensible world raises the question of the relation between freedom and autonomy. Kant seems to identify the two, when he asks, rhetorically: “what, then, can freedom of the will be other than autonomy?” It is relatively easy to see how I might act freely in the intelligible world, imposing a character upon myself as belonging to the sensible world. And insofar as character serves as a kind of law for my conduct in this world, I can easily be understood as autonomous—as giving myself a law. But it is not obvious, then, why Kant would explain autonomy as our giving ourselves the moral law. The law of a self-imposed sensible character seems to be one thing, the self-imposed moral law seems to be another. It is important to notice here, however, that sensible character includes subjection to the moral law. For this reason, to self-impose a character is to subject myself to the moral law. How I react in the sensible world to the recognition that my contemplated action would be morally wrong is determined by my character; and the only way this can be true is if my character includes some degree of respect for the moral law. Hence, to impose upon myself a law of action (character) for the sensible world, through a free act in the intelligible world, is to subject myself to the moral law. Freedom of the will and autonomy are thus, in this sense, roughly the same.



Hill, Thomas E., Jr. (1992), Dignity and Practical Reason in Kant’s Moral Theory (Ithaca, NY: Cornell University Press).

Kant, Immanuel (1997a), Groundwork of the Metaphysics of Morals, trans. Mary Gregor (Cambridge: Cambridge University Press).

_____ (1997b), Critique of Practical Reason, trans. Mary Gregor (Cambridge: Cambridge University Press).

_____ (1998), Critique of Pure Reason, trans. Paul Guyer and Allen W. Wood (Cambridge: Cambridge University Press).

Korsgaard, Christine M. (2009), Self-Constitution: Agency, Integrity, Identity (Oxford: Oxford University Press).

McCarty, Richard (2009), Kant’s Theory of Action, (Oxford: Oxford University Press).


© Richard McCarty
Department of Philosophy & Religious Studies
East Carolina University

Last modified: August 28, 2015

Terms of use.  Access to this copyrighted electronic document is open to everyone. Permission is hereby granted to quote any parts under 500 words, provided the author’s name is cited, along with the names and sources of any other authors quoted. Permission is also granted to quote any parts under 500 words in translation, provided the author’s name is cited, along with the names and sources of any other authors quoted. Duplicating this copyrighted material without obtaining prior, written permission is allowed only for local instructional use (“fair use”), which must include this statement of terms of use. Plagiarism is not allowed. Your professor can find this resource as easily as you did.