2.1 The Lawless Will: A “Nonentity”

To understand why Kant views freedom as a kind of causality, it is helpful first to consider how he understands the faculty of will as such. This brings us to Section II of the Groundwork, where we find Kant defining the will on three separate occasions:

As we can see, each definition builds upon the former, yielding a more comprehensive view of what it means to will. The first definition, or what I am calling will-1, is the most minimal of the three inasmuch as it says merely that the faculty of will is a capacity to act according to the representation of laws. Will-2 goes further by clarifying the mode of acting in terms of self-determination. Finally, will-3 reveals the kind of laws that are left unstated in the second definition, specifying them in terms of the will’s own capacity for universal legislation. A complete analysis of the faculty of will thus yields insight into what Kant calls the supreme principle of morality, namely, the principle of autonomy. As a rule, it states the most general formula of moral action, to the effect that one can, in the maxims one adopts, consider oneself at the same time as giving universal legislation (GMS, AA 4:434). That is the thrust of Kant’s argument in Section II: that what morality commands of us is “neither more nor less than just this autonomy” (GMS, AA 4:440).Footnote ¹³

Given this conclusion, it is not surprising that Kant returns in Section III of the Groundwork to the concept of causality. “A will,” he writes, “is a kind of causality of living beings in so far as they are rational” (GMS, AA 4:446). At first Kant adds that “freedom would be that property of such a causality as it can be efficient independently of alien causes determining it” (GMS, AA 4:446) – a definition Kant admits is of little value.Footnote ¹⁴ But he does not stop here, for Kant thinks we can move deeper into the concept of a will and attain a positive grasp of its freedom (as a definition). This brings us to the doctrine of moral necessitarianism, the claim that the moral law is the causal law of a free will. Kant argues as follows:

This is a complex passage, to be sure, but for our purposes what matters is premise 2, since it is only by denying the possibility of a lawless will that Kant can secure his inference to the claim that a will must have a causal law. The claim that the principle of autonomy is the only suitable candidate for the will’s law follows straightforwardly, since all Kant needs to show is that a law exhibiting “heteronomy” (such as a law of natural mechanism) is antithetical to the will’s self-determination. For this reason many commentators have found it puzzling that the key premise in Kant’s doctrine of moral necessitarianism appears without argument, which to the defender of leeway freedom and moral liberty begs the question. Even if we grant some version of Kant’s transcendental idealism and assert that the law of natural mechanism applies only to the world of appearances, why should we think that our noumenal agency is subject to moral necessity? Is it not preferable to think of our phenomenal selves as determined, but our noumenal selves as unconstrained by all rules?

Two initial considerations might speak on behalf of moral libertarianism. The first is that it captures a popular intuition that to be free is to be unconstrained by rules of any kind. In this sense of the term, my freedom consists in my spontaneity, and the more spontaneous I am, the freer I am. We could even say that my freedom is lawless, for there is nothing to stop me from, say, randomly throwing large sums of money into a fire, bursting into song during a lecture, or doing cartwheels down the streets of Toronto. I need not provide reasons for why I did such things, but that does not mean that in doing them I was unfree; indeed, one might think I was freer, since my deeds were unconstrained by any reason. Suppose I had good reason not to throw large sums of money into the fire, but by nevertheless doing so, I proved my freedom – my capacity to rise above all norms of custom or morality or even rationality.Footnote ¹⁵ On this picture, Kant’s claim that a lawless will is a “nonentity” (Unding) appears counterintuitive. At the very least, he could have said just that a lawless will is an imprudent will, but instead he made the stronger claim that the concept is self-contradictory.

Another reason we might want to call moral necessitarianism into question is that it creates a potential problem for the imputation of evil to an agent, which itself is a variation of the Objection from Necessity outlined earlier. Reinhold was one of the first readers of Kant to voice this concern, according to which Kant’s identification of morality and freedom undermines the imputation of immoral action:

Reinhold did not think Kant was guilty of creating this problem, though he worried that the Groundwork and the second Critique lend themselves to this misunderstanding. For this reason Reinhold redefines freedom as the ability to act without being constrained either by the laws of reason or by the demands of sensibility, adding that someone exhibits free “choice” (Willkür) by deciding to “determine himself” by reason or to “let himself be determined” by objects of sensibility (1789, 90). He reiterates this point in the second edition of his Letters on Kantian Philosophy (1792), claiming that the will is “the capacity of a person to determine itself to the satisfaction or nonsatisfaction of a desire, either according to the practical law or against it.” Reinhold thereby defines freedom as liberty of indifference, or the capacity of choice to act either “for or against the practical law” (1792, 271–272). We are responsible for an evil character, he claims, just as we are responsible for a good character because we are at liberty to choose one or the other.

2.2 The Principle of Determinate Lawfulness

In light of Reinhold’s worry, it is natural to ask why Kant ever seemed attracted to moral necessitarianism in the first place. To address this question, we must return to premise 2 of Kant’s argument in Section III of the Groundwork (quoted earlier) and try to clarify the reasons he had for upholding it. Upon inspection, it is clear that Kant’s rejection of the possibility of a lawless will is itself an inference from an unstated principle in premise 1. Recall that premise 2 states that a free will must be a “causality according to immutable laws.” Kant later distinguishes between two kinds of laws: those that operate from without (heteronomously) and those that operate from within (autonomously). These remarks follow from what Kant establishes at the beginning of his argument, where he defines the concept of causality in terms of a law that for any cause, a consequence must be posited (see GMS, AA 4:446). What is noteworthy is that Kant has already stated a version of this principle in Section II, that “everything in nature works according to laws” – or what we might call the Principle of Determinate Lawfulness:

When Kant formulated this principle earlier in the Groundwork, he did so in the context of defining the will: “Only a rational being has the capacity to act according to the representation of laws, i.e. according to principles, or a will” (GMS, AA 4:412). This definition of will-1 and its clarification in what I have termed will-2 and will-3 are crucial steps in Kant’s derivation of the supreme principle of morality. But it is only when we arrive at his argument that the moral law is the causal law of a free will in Section III that these earlier definitions come back into play. For Kant’s larger aim is to show that, given the Principle of Determinate Lawfulness, nothing happens without a causal law – a point he makes clearly in the first Critique: that “every effective cause must have a character, i.e., a law of its causality, without which it would not be a cause at all” (A538/B566). This is the version of the Principle of Sufficient Reason that Kant held during much of his pre-critical period: that for everything that happens or ought to happen – including “nature” in all its domains – there must be a character or law of its occurrence.

The application of this principle becomes clear when we consider actions or events in the phenomenal world. When we say, “At sea level water boils at 100°C,” we are not predicting the outcome based on repeated past experiences of water boiling at this temperature. Rather, the judgment is about an objective synthesis of appearances that operates according to a rule; that is why, barring impurities of water or changes of pressure, water at sea level always boils at 100°C. One event necessarily follows from the other. Moreover, as early as 1755 Kant recognized that for every action or event, all other alternatives are excluded. The boiling point of water operates according to a fixed causal rule, which means that all other possible boiling points are excluded from this phenomenon. And this makes sense, given that we would not have a unified experience of the world if water at sea level boiled at 100°C on some days, at 72°C on other days, and at 97°C on still other days. For Kant, the regularity of phenomena is due to the necessity of rules making up our ongoing apprehension of a manifold, which we call the sensible world.

Consider next the application of this principle to the case of a perfectly good agent, or what Kant calls a “holy” will (GMS, AA 4:414). Let us imagine a will unaffected by desires, impulses, or any of the tendencies of self-deception that Kant attributes to the common psychology of human beings. The task is to represent an agent who experiences nothing within herself that could tempt her to question the validity of moral laws. For this special individual, her agency would consist in always acting in accordance with her autonomy: She would always choose maxims that she sees as expressing her standing as a universal lawgiver, while recognizing at the same time that this status puts her on an equal footing with all other members of the moral community. For such an agent, whose will is perfectly good, what she recognizes to be required by the laws of morality would immediately exclude alternate possibilities: a duty to a keep a promise, for example, would strike her as a rule that closes off its violation, including any possible course of action that fails to respect this duty.Footnote ¹⁷

My point in introducing this idealized agent is to foreground a symmetry between how the Principle of Determinate Lawfulness applies in the theoretical domain, where we apprehend phenomena in a manifold of experience, and how it applies in the practical domain, where we formulate maxims in a manifold of willing.Footnote ¹⁸ The character of natural phenomena (such as water boiling at 100°C) is a causal rule that explains why things happen; the character of normative phenomena (such as a duty to keep a promise) is a causal rule that explains why things ought to happen. In the manifold that we call the sensible world, rules operate to the exclusion of alternate possibilities, and that exclusion is what makes a determinate event of experience possible. Similarly, in the manifold we call the moral world, rules operate to the exclusion of alternate possibilities too, and that exclusion is what makes a determinate act of willing possible. To say that the moral law is the causal law of a free will, then, is to say that the moral law is the rule of the will’s self-determination, which as a principle excludes all forms of willing that are subject to external influences.

This helps to bring out the argumentative force of premise 2, if only because it dispels the impression that a will “under moral laws” is somehow unfree. Consider again our idealized agent who is conscious of her autonomy and only ever forms maxims that express her recognition of this autonomy. We can imagine her feeling elevated by the awareness that she is a universal legislator subject to no external authority – not even God’s – because she is living in accordance with her own faculty of reason. If asked, she would likely say that she feels most free whenever she acts under moral laws, because she sees herself at the same time as a legislator of those laws. Considered from this standpoint, the rest of humanity – all of us who pursue our desires at the expense of our duties – appear to be so many “slaves of passion” who are less (not more) free for acting randomly or doing whatever we wish. For the idealized agent, the exclusion of alternate possibilities demanded by the moral law is not a negative restriction: rather, it is a condition of her will’s perfection, that of being an independent, spontaneous, and self-legislating power.

Kant’s further point is that what makes the will of a finite rational being different from natural phenomena like boiling water is that the will is a causal power that admits of degrees of perfection. The conformity of the will to its inner law – the moral law – is a special kind of normative restriction that corresponds to the will’s elevation because it is self-imposed. Thus, the person who feels more free the more she acts in accordance with the laws of morality is not deceived, in Kant’s view. For she correctly feels that her self-imposed restriction under moral laws is a perfection of her will as a causal power, one that in principle excludes influences – even those stemming from her sensibility – that undermine her autonomy. In this sense, acting according to moral laws does exclude alternate possibilities, and to that extent it limits rather than expands one’s leeway liberty, but not in a way that decreases one’s freedom. If we represent to ourselves the upper limit of the will’s perfection, as in the case of a holy will, we can see that there is no freedom to do otherwise because such a will excludes nonmoral courses of action, and yet its freedom is still absolute.

Thus the Principle of Determinate Lawfulness is the unstated basis for Premise 1, which led Kant to say that a lawless will is an absurdity. The will, like everything in nature, must have a law of its causality or character, “without which it would not be a cause at all.” That law cannot be a principle of natural causality because such causality operates from without, conflicting with the will’s self-determination. The only law left is the principle that the will is a law to itself, which is the moral law qua principle of autonomy. Hence, the character of a free will is that of acting on maxims it can also recognize as universally legislating. And that is why the will is a causal capacity to initiate action “from itself” (von selbst): its restriction to the moral law is at the same time an expansion and elevation of its power. The more a finite rational being approximates the ideal of such power – the ideal of “holiness” – the more it excludes actions that are contrary to morality. As Kant explains, there is no power to do otherwise built into the definition of freedom; acting contrary to the law is not a Vermögen (an ability or power) but an Unvermögen (MS, AA 6:227).Footnote ¹⁹

Another way to unpack this line of argument is to think of lawfulness in terms of unity.Footnote ²⁰ For Kant, every power of the mind has a supreme principle, a governing law of its use, and he is clear that this law is what makes each power both unifying and unified. Sensibility is a power of receiving a manifold of empirical intuitions, and its supreme principle is to bring those intuitions under the unity of space and time. Understanding is a power of combining such a manifold, and its supreme principle is to bring it under the “unity of apperception” (B136). Reason is a power too, and for Kant its legitimate use is to direct all faculties toward “architectonic unity,” whereby all parts of human cognition are “purposively united with each other as members of a whole in a system of human cognition” (A835/B863; cf. A298/B355). For Kant, on my reading, the power of willing is no different: the more we perfect the power, the more we bring our maxims under the unity of moral laws.

2.3 Noumenal Determinism and Ulrich’s Fork

Nevertheless, the initial worries behind the Freiheitsstreit still remain with us. As mentioned in §1, the point that Ulrich wanted to make in his Eleutheriology (Reference Ulrich and Heinrich1788) is that, rather than protecting freedom from the forces of necessity, Kant has only reintroduced those forces by making the will subject to the moral law’s causality at the noumenal level. Kant is committed to making natural necessity rule without exception over the world of appearances, he explains, including the empirical character of our will. Yet the empirical character of our will is, by Kant’s lights, grounded in an intelligible character whose causality must operate according to laws as well. According to Ulrich, we have “necessity with the immutability of the intelligible character (albeit not natural necessity in the Kantian sense),” the result being that “necessity reigns here too” (1788, 32). Nor does Ulrich believe that Kant avoids the problem by redefining freedom as the power to choose otherwise, that is, liberty of indifference, since that would raise a problem of “indeterminism.” Because a power to choose otherwise is a power without “decisive reasons,” the will is thereby reduced to mere “chance” (Reference Ulrich and Heinrich1788, 21).

At the heart of Ulrich’s criticism is the claim that between necessity and chance there is no third alternative. This we might call Ulrich’s Fork:

Looking back, we can see how this fork might be a problem for Kant’s Principle of Determinate Lawfulness. Recall that our discussion in §2.2 was meant to show why being restricted to a rule does not necessarily limit a power. Restriction to a rule of natural causality is what makes my cognition of actions or events in a sensible manifold determinate, and the application of that rule to specific phenomena is what allows for a unified experience of the world. Working toward systematic unity of cognition under the rule of reason is also a progressive perfection of my mind. However, restriction to that rule does not limit my mind: conformity to the law of any cognitive faculty is what gives that faculty unity, and its progressive perfection, for Kant, is what enhances its power – it does not diminish it. And yet, to play devil’s advocate on Ulrich’s behalf, we might point to an asymmetry between the theoretical and practical domains. This is because, however active the mind is in its cognitive operations, only a pure will exhibits what Kant calls “absolute spontaneity,” and it is here that symmetry between the two domains no longer holds.

As it turns out, Kant does not need perfect symmetry between the theoretical and practical domains to make his argument work. In various texts he is explicit about the fact that spontaneity of sensibility and spontaneity of understanding are both relative, since they both display a degree of dependence on a given manifold of intuition (see A546/B574; GMS, AA 4:452). Reason is a unique power in this regard, since its operations are elevated above sensible intuition; and the “cosmological” freedom of reason in its pure practical use – the pure will – is the most elevated of all, since it is a causal power to initiate action von selbst, as Kant says (A446/B474). My point in drawing parallels between these applications of reason is only to show that everything that happens or ought to happen stands under a causal rule, as its character or law.Footnote ²² While some might find it attractive to argue from the spontaneity of theoretical cognition to freedom of will, this is not an option for Kant, especially since he believes that it is only consciousness of the moral law that “discloses” our freedom to us (see KrV, AA 5:29–30).Footnote ²³ The relevant point here is that only the connection between moral law and will is the right kind of causal connection for autonomy; all else counts as “heteronomy.”

Turning the tables now on Ulrich, we can say that the Objection from Necessity rests on an equivocation between two distinct senses of “causal connection.” For simplicity, let us call an autonomous causal connection “a-causal” and its corollary necessity “a-necessity,” and call a heteronomous causal connection “h-causal” and its corollary necessity “h-necessity.” With these distinctions in place, the premises of the Objection from Necessity are more precise:

1. (1) (moral necessitarianism) The moral law is the necessary a-causal law of a free will.

2. (2) A will subject to the h-causality of the moral law is determined.

3. (3) (noumenal determinism) Therefore, a moral will’s activity reduces to h-necessity.

The slide in meaning from premise 1 to premise 2 makes all the difference. For once we assume that the will is subject to a heteronomous causal law (h-causality), it follows that the will’s activity reduces to heteronomous necessity (h-necessity), leaving us with the problem of noumenal determinism. However, the original version of the Objection from Necessity begs the question to the defender of a-causality and a-necessity. Nothing about the argument demonstrates that such causality/necessity is impossible; instead, it assumes that all causality/necessity must be of a heteronomous (or mechanistic) variety.Footnote ²⁴

To avoid the objection, all we need to do is write the premises consistently:

1. (1) (moral necessitarianism) The moral law is the necessary a-causal law of a free will.

2. (2*) A will subject to the a-causality of the moral law is determined.

3. (3*) (no noumenal determinism) Therefore, a moral will’s activity reduces to a-necessity.

Now this raises the question: What are a-causality and a-necessity? Everything turns on this, since Ulrich’s Fork – the claim that there is “no middle way between necessity and chance” – is a problem for Kant’s theory of freedom only if by “necessity” he means h-necessity exclusively. To preempt the Fork, all we need to do is establish grounds for distinguishing autonomous causal connections from heteronomous ones. If we can do this, securing premise 2*, we can show that the a-causality and a-necessity of a moral will enhance the will’s freedom, securing conclusion 3*. This would yield an important payoff – that Kant’s moral necessitarianism does not entail noumenal determinism after all.

2.4 Necessitation and Causality

While Kant has left many resources to secure premise 2*, the most relevant concerns a pair of distinctions he often draws in his critical work: (a) the distinction between practical necessitation and natural necessitation, and (b) the distinction between noumena and phenomena.Footnote ²⁵

To begin with, we can understand necessitation as such in line with the claim that everything in nature operates according to laws – what I have termed the Principle of Determinate Lawfulness. Necessitation will then refer to the way laws govern things, namely, by excluding alternate possibilities and applying to all cases of a kind. Necessitation is “natural,” according to Kant, (i) when the causal connection lies between two appearances qua phenomena, and (ii) when the causal connection lies between two appearances in time. The time condition is already contained in the first definition, but it serves to bring out the following rule of natural necessitation:

This highlights a distinctive feature of natural causality and its form of necessitation. In terms of time relations, all effects are preceded by an infinite series of antecedent conditions, since for every cause of an effect (as the rule states) there must be a prior cause, which in turn must have a prior cause, and so on ad infinitum. As Kant puts it in the first Critique, the unrestricted version of the law of natural causality rules out the possibility of cosmological or transcendental freedom, the power of beginning a new series “from itself,” since all causal connections in nature must be preceded by antecedent states in time. This is why Kant adds that for the “physiocrat” – the person who believes that “all is nature” in this narrow sense of the term – freedom is a nonentity, an “illusion of the mind” (A449/B477).

Before turning to this objection, let us consider what the concept of nonnatural necessitation contains. We now know that necessitation as such refers only to the way laws govern things (by excluding alternate possibilities and applying to all cases of its kind). If natural necessitation occurs (i) when the causal connection lies between two appearances (qua phenomena), and (ii) when the causal connection lies between two appearances in time, what might nonnatural necessitation look like? As a matter of definition, it must be a necessitation that does not lie between phenomenal appearances in time. So what might that be? Kant’s answer is that practical necessitation is still a species of causality, but one that operates through concepts, where the representation of concepts provides an agent with ends she ought to realize through her willing.Footnote ²⁶ Thus the structure of practical necessitation is similar to that of natural necessitation, since it operates through a twin process of determination and exclusion of alternatives:

This distinction lies behind what Kant says in the first Critique about the two ways we can think of causality in general. “In respect of what happens,” he writes, “one can think of causality in only two ways: either according to nature or from freedom,” to which he adds:

Kant’s further point is that nonnatural causality is conceivable only at the level of things-in-themselves, since all relations between appearances trigger an infinite regress that forecloses any causal power beginning from itself. That is why the rule of Practical Necessitation makes no reference to time determinations. For any given maxim M, there must be a cause C, and that C must be antecedent in the order of reason, and for Kant this must be an intelligible order rather than a temporal one. His example of getting up from a chair is helpful to consider here: for when I get up from a chair, I initiate an “absolutely new series, even though as far as time is concerned this occurrence is only the continuation of a previous series” (A450/B478). From the standpoint of appearances, the event of getting up is tied to a chain of causes extending back in time infinitely. From an intelligible standpoint, by contrast, my act does not lie “within the succession of merely natural effects and is not a mere continuation of them,” but counts rather as an “absolutely first beginning of a series of appearances” (A450/B478; cf. A553/B581). Yet this makes sense only if we distinguish between these two domains, the phenomenal and the noumenal.Footnote ²⁸

This helps to clarify the nature of the autonomous causality (or a-causality) that underpins the doctrine of moral necessitarianism. As a matter of definition, a-causality cannot be a causality that operates at the level of appearances, since that would reduce to heteronomy, with each event being preceded by antecedent events in time. More positively stated, and still as a matter of definition, a-causality must operate at the intelligible level of concepts, which leaves open the possibility of transcendental freedom, the power to initiate a new series from itself. The reason why a-causality can have only one law is because there is only one law consistent with the will’s independence from external influences. As we know, the supreme principle of morality just is the principle of autonomy, which states as a rule that the will can always recognize in its maxims its own giving of universal legislation. There is only one principle consistent with the idea of a self-generating causal power, and that is the moral law.Footnote ²⁹

Remember too that for Ulrich’s Fork to pose a real problem, it must be true that there is an exclusive disjunction between heteronomous necessity on the one hand and chance on the other. But there is no such disjunction, since it is conceivable to represent a form of necessity that results from nonnatural (and so nonheteronomous) causality: this is the necessity of a will acting according to the principle of its own universal legislation, whereby its activity is from itself and so fully spontaneous. Pace Ulrich, the claim that the moral law is the causal law of a free will does not entail noumenal determinism, since the causality in question is consistent with the will’s independence. This allows us to say, having established premise 2*, that a will subject to the a-causality of the moral law is “determined” (3*), without the determination in question undermining its freedom; rather, this is the kind of self-determination that marks the perfection of the will, not its destruction. A will that restricts itself to the same principle it gives to itself is not less free, but more free.Footnote ³⁰

The idea of an intelligible ground then allows us to use the category of causality in new ways. From a theoretical standpoint, this category applies only to the sensible domain of cognition; Kant’s point, however, is that this same category acquires new significance in connection with the intelligible domain of volition. And this is worth emphasizing, as it explains the contrast between the concept of causality according to nature and that of “causality from freedom.”Footnote ³¹ Since the order of priority that makes up the space of an intelligible ground can refer only to an order of reason and not to an order of time, the causality of the moral law can refer only to a normative relationship between a “ground,” the idea of a self-legislating Wille, and its “consequence” for our power of Willkür, namely, to act in accordance with the law.Footnote ³² Any cause-effect relation that pertains to the phenomenal world is mechanistic – and so antithetical to freedom – since it mandates that for every action or event in time there must be a prior action or event. For Kant, the space opened up by transcendental idealism allows us to redeploy this category to show how a power of absolute spontaneity beginning from itself can still count as a lawful power.

Having reached this conclusion, it is worth pausing to reflect on where we stand in the dialectic of the Freiheitsstreit. As we have seen, the controversy developed into the form of the Free Will Dilemma, forcing Kant to define freedom of will in a way that reduces it either to necessity (the first horn) or to blind chance (the second horn). Ulrich was the first critic who argued that Kant is guilty of reintroducing the problem of determinism at the noumenal level by defining the moral law as the causal law of a free will. I have attempted to preempt this criticism by showing that it turns on an equivocation between two senses of causal necessity; for once we keep this distinction in view, the Objection from Necessity loses its force. But then even if this is true, what are we to make of the second horn of the dilemma? And what can we say about Kant’s apparent acceptance of a liberty of indifference view in his 1792 essay “On the Radical Evil of Human Nature”? Is this one place where Kant leaves himself exposed to the Objection from Chance?

In §2.5 I will argue that it does not. Kant understands that the Objection from Chance is a problem for theories of leeway freedom and moral liberty, I believe, and that is why he is consistently committed to source freedom and moral necessity.

2.5 Responsibility for Evil

There are few places in Kant’s published or unpublished work where he defines freedom of will as the ability to do otherwise, as the ability to choose for or against the moral law, or any similar formulation. This is not because philosophers of his day lacked the terminology to describe this position: The phrase libertas indifferentiae was in wide circulation well before the eighteenth century, and it was central to the work of Kant’s older contemporary, Christian August Crusius (1715–1775). In one of Kant’s first publications, the New Elucidation of 1755, we even find a mock dialogue between a defender of determinate grounds (Kant’s position at the time) and a defender of leeway freedom (the position of Crusius and his followers), and it is instructive to see how he frames the thesis of leeway freedom:

In reply, Kant’s interlocutor accuses this proponent of leeway freedom of ruling out any explanation of why the agent acts one way rather than the other, thereby reducing the will’s activity to a “product of chance” (PND, AA 1:402). Curiously, this line of criticism reappears in Creuzer’s Skeptical Observations on the Freedom of the Will (1793). By defining freedom as the capacity to choose for or against the moral law, Creuzer upholds, we must think of the will as subject to two ruling forces, at one time reason, at another time sensibility, without having any determining ground to explain what compels the agent to choose one over the other (128 n).Footnote ³³ If Kant did read Creuzer’s book, he would have had a feeling of déjà vu, for this was the same objection he had raised against Crusius thirty-five years earlier. Now, in a strange turn of events, it was being leveled against Kant himself.

The issue here is not merely interpretive, however. Even if we evade the Objection from Chance by showing on textual grounds that Kant was never committed to leeway freedom, we must still address a conceptual worry surrounding his doctrine of moral necessitarianism (what I have termed the Problem of Imputation). To recall, Reinhold was concerned that defining the moral law as the causal law of a free will would render immoral action unfree, which would make it impossible for us to impute evil to a rational being: an unfree agent can be no more responsible for evil than a natural disaster is. This does not fit with Kant’s repeated insistence that human beings are responsible for both their moral and immoral deeds. On Reinhold’s view (as we saw in §2.1), the only solution is to characterize freedom as the power to choose for or against the moral law. But if Kant never adopted that view, as I am claiming, then what could he say in response to the Problem of Imputation?

As a preliminary step, it is important to see that Reinhold’s account rests on the following assumption:

Reinhold himself never argued directly for this claim, perhaps because he thought it was self-evident that leeway freedom is the only kind of freedom that can support moral imputation. Yet we have reason to question this, quite apart from how we interpret Kant, since it raises anew the specter of chance. If there is no determinate ground for my willing one way rather than the other, then it seems that my willing is a product of blind chance. Moreover, it is no help to the libertarian’s case to raise this exercise of will to the noumenal level, since noumenal chance would be just as problematic as chance at the phenomenal level. I say this only to emphasize that the Problem of Imputation on its own does not mandate a commitment to leeway freedom or moral liberty. If moral libertarianism threatens to undermine the concept of freedom, then it faces an even worse version of the imputation problem, since it potentially renders both moral and immoral action mysterious.Footnote ³⁴

Arguably, the doctrine of moral necessitarianism faces a less daunting challenge, for it has only to explain how immoral action counts as free action. I believe Kant left enough hints to show why a commitment to source freedom and moral necessity does not rule this out a priori. The first point to note is that Kant never equates (1) resistance to acting according to the moral law with (2) a genuine Vermögen (power or ability). Nor, relatedly, does he ever say that finite rational beings can have an incentive to do evil for the sake of evil, or – which amounts to the same thing – that they can have a reason to violate the moral law for the sake of violating the moral law. Kant makes these points with such frequency that their import can sometimes be lost; but they are key aspects of his moral necessitarianism. To say that the moral law is the causal law of a free will means that it is the supreme norm for the will’s unity. It is supreme, for Kant, because it is the only law consistent with the idea of autonomy, the only principle according to which I can regard myself at the same time as a universal lawgiver.

Putting these claims together, it becomes clear why Kant says that a “diabolical” agent is a contradictory concept (RGV, AA 6:35).Footnote ³⁵ A diabolical agent, if such an agent were possible, would be someone who freely acts against the moral law, not because of a random whim or inclination, but because he has an incentive for such a violation; in short, he is someone who wills evil for the sake of evil. But this means that he is someone whose exercise of free will operates against his own free will: the very condition of his choice of maxims, the supreme norm of their unity, is not made the object of self-deception or exception-making (as Kant believes is the case for ordinary human evilFootnote ³⁶). Rather, we are to imagine an agent for whom acting against the condition of his own acting is itself somehow a reason for acting. Inasmuch as this is a contradiction of concepts, a diabolical will is not possible. And this supplies further evidence for Kant’s commitment to the thesis that the moral law never stops operating as a “causal law” for finite rational beings, since it never loses its place of authority as the will’s highest determining ground.

But then the question becomes: Are not those self-exceptions that make up the scope of human evil freely chosen? And are they not chosen with at least some degree of awareness of the moral law’s authority? If so, it would appear that even those more mundane forms of human evil are performed contrary to the moral law, and hence are “unfree” by the lights of moral necessitarianism. In reply, I think this conclusion is too quick, since it does not recognize the fact that a causal law of a power can coexist with a misuse of that power. All the doctrine of moral necessitarianism states is that the moral law is the necessary a-causal law of a free will. This means that a power to initiate a new series from one’s will is possible only under the rule of one’s autonomy; thus the moral law is the only principle that makes the will a transcendental power and cosmological ability. However, the doctrine does not rule out the internal diminishing of this power through weakness, or its corruption through impurity, or even a perversion of it through self-conceit. All the grades of the “propensity to evil” we find listed in the Religion are so many ways of misusing freedom, but none of them amounts to a power or ability.Footnote ³⁷

On the contrary, these grades of the propensity to evil – weakness, impurity, self-conceit – are so many grades in the imperfection of the will, and for Kant an imperfection always disempowers a faculty, never the reverse.Footnote ³⁸ Nor do these grades of imperfection unseat the moral law from its place of authority in the mind of a rational agent, for Kant believed that human immorality always occurs as a form of self-deception or exception-making. We never pursue evil for the sake of evil, but we do try to work our way around the moral law, by questioning its strictness, its purity, or its validity. Because the moral law is the causal law of a free will, we cannot renounce it without, per impossibile, renouncing our freedom. These cases of moral failure, considered as so many degrees of imperfection, are not instances of a Vermögen at all. Given that they do occur, Kant is committed to saying that they are possible (since actuality entails possibility)Footnote ³⁹ – but at no point does he characterize the choice of an evil character as a real ability, nor does he define freedom in terms of an ability to do otherwise.Footnote ⁴⁰

One reason why Kant’s readers have at times thought he was committed to moral libertarianism is that his appeal to the actuality of human evil always comes with the admonition that “we ought nevertheless to be good.” Such an “ought” claim seems to imply that in those situations where we have given in to weakness, impurity, or self-conceit, we could have done otherwise, on the grounds that “ought implies can.” But this conclusion, however self-evident it might appear, rests on a false inference. It is correct to say that when we ought to have acted according to the moral law we could have done so; but this does not entail that at the moment when we failed to act according to the moral law, we enjoyed that place of which the moral libertarians speak: the place where we stand indifferently between two or more courses of action. For Kant, there is no such place – it is an Unding. We are not in fact indifferent to the moral law; its commands resound “unabated in our soul,” as Kant says (RGV, AA 6: 45), and those commands do not go away when we fail to exercise our power of freedom. Even in those instances of moral failure that we call evil, Kant claims, the moral law remains within us.Footnote ⁴¹

In my view, this is not a departure from Kant’s account in the Religion; rather, it sheds light on what he was trying to defend throughout much of his career. As he says in the Metaphysics of Morals:

One might think that this conflicts with Kant’s argument in the Religion that for evil to be imputed to an agent, it must be ascribed to his freedom of choice at the noumenal level – specifically, in his adoption of an evil “disposition.” On my reading, there is no tension here, for in the Religion Kant only ever argues from the empirical fact that human beings can and do act immorally, to the condition of the possibility of acting immorally. His point is that because this condition cannot emerge from a natural series of time determinations – for then it would not be free – we must ascribe it to a human being’s intelligible character,Footnote ⁴² which is not subject to such time determinations.Footnote ⁴³

In drawing this inference, all Kant is saying is that the evil we witness in the world must have its source in our freedom. He is not making the further inference that scholars sometimes draw on his behalf: that we have a genuine power to choose for or against the moral law at the noumenal level. For Kant, ascribing liberty of indifference to an agent’s will would commit the fallacy of defining a thing-in-itself through a thing-as-it-appears. Certainly, experience provides examples of such indifference, but the question is only whether we have grounds to define freedom in such terms. On my reading, this is what Kant wants to avoid. “For it is one thing,” he writes, “to accept a proposition (on the basis of experience) and another thing to make it the expository principle (of the concept of free choice) and the universal feature for distinguishing it” (MS, AA 6:227). In the Religion, when Kant ascribes evil to an agent’s intelligible character, he is careful to speak of it as a corruption of an agent’s original predisposition to the good, adding that it is a “mystery” we cannot comprehend.Footnote ⁴⁴ To say that evil is the result of an ability to do otherwise would be, by Kant’s lights, to overstep the bounds of what we can know.Footnote ⁴⁵

When Kant refers to evil, moreover, he is careful to gloss “corruption” (Verderbtheit) with the Latin corruptio (RGV, AA 6:30); in this way he places his discussion within a broadly Augustinian framework for thinking about evil as a deprivatio, negatio, or defectus.Footnote ⁴⁶ Though we cannot conclude from Kant’s choice of terms here that he is alluding specifically to Augustine,Footnote ⁴⁷ it does give us a clue for understanding how a misuse of our ability of choice is a corruption of it. The idea of evil as a corruption of choice speaks to why Kant rejects liberty of indifference, since a corruptio is strictly speaking the removal or destruction of something; thus, to corrupt one’s power of choice by choosing evil is to be deprived of that power. So when Kant said that the ground of evil cannot be located in the “corruption of the morally legislative reason,” his point is that the causality of freedom cannot “uncause” itself. That would be “as if reason could extirpate within itself the dignity of the law itself,” which Kant argues is “absolutely impossible.” To regard oneself “as a freely acting being, yet as exempted from the one law commensurate to such a being (the moral law), would amount to the thought of a cause operating without any law at all, … and that is a contradiction” (RGV, AA 6:35).Footnote ⁴⁸

3.1 The Controversy Continues

In the spring of 1793, Kant received the following letter:

This letter is dated April 7, 1793, and its author was none other than Leonhard Creuzer, then twenty-five years old, who had just published his “little book,” Skeptical Observations on the Freedom of the Will. We can surmise from the fact that Kant received this letter that he was aware of the Freiheitsstreit his theory had set in motion. However, there is no evidence that Kant replied to Creuzer, or that he ever read his book.Footnote ⁴⁹ We can only speculate what Kant would have thought about the Objection from Chance that Creuzer wanted to raise against theories espousing liberty of indifference. As it happens, someone close to Kant’s philosophy took on the task of writing a review of Creuzer’s study for the Allgemeine Literatur-Zeitung, which appeared in the October 1793 issue. The review was left unsigned, but the reading public would soon learn more about its author, Johann Gottlieb Fichte (1762–1814), to whom we shall return below.

Kant’s awareness of the Freiheitsstreit is also evinced in a letter he received from Salomon Maimon (1753–1800) on November 30, 1792 (letter 548, AA 11:389–393).Footnote ⁵⁰ The letter was composed soon after Reinhold published the second edition of his Letters on Kantian Philosophy, which included a new discussion of the free will problem in the eighth letter, entitled “Exposition of the Concept of the Freedom of the Will.” Surprised by Reinhold’s commitment to moral libertarianism, Maimon wrote to Kant that Reinhold’s concept of a free will “leads to the most inexplicable indeterminism” (AA 11: 390). As he explains, Kant’s theory of freedom involves a “causality of reason,” and yet on Reinhold’s alternative “the causality of reason would be a natural necessity” (AA 11: 390). Further, he argues that on Reinhold’s view of freedom as the capacity to do otherwise, either for or against the moral law, the resulting definition does not “in the least” concern itself with the “determining ground” of freedom (AA 11: 390). In other words, Maimon was voicing a version of the worry Kant had long felt regarding the notion of liberty of indifference: that it runs afoul of the Principle of Determinate Lawfulness. Maimon perceived this as well from the standpoint of his commitment to the Principle of Sufficient Reason, and he was keen to know what Kant thought of Reinhold bringing his theory under the banner of indeterminism.Footnote ⁵¹

Maimon went on to pose this question in a series of letters he wrote to Reinhold himself, later published in a collection titled Quarrels in the Field of Philosophy (1793), asking how the will can decide between two opposed drives and what reason it can give for selecting one over the other. “The question,” he states, is why the will “in one moment follows the instruction of reason, and in another moment the selfish drive” (233). Maimon confesses that he cannot understand this, and this leads him to formulate a version of the Free Will Dilemma according to which the Reinholdian will reduces to either natural necessity or blind chance:

As this makes clear, the source of the dilemma is Reinhold’s brand of leeway freedom, his claim that the will can choose between the operations of reason and sensibility. For Maimon, as soon as we accept such indeterminism about the will, there is no way to explain how the will can determine itself to act at all. The concept of an indifferent will is thus an “empty” concept, since by violating the Principle of Determinate Lawfulness, it amounts to what Maimon called a “powerless power,” which is simply a figment of thought.

Reinhold for his part did not receive these criticisms gratefully:

That Reinhold frames his question in these terms is revealing. For it shows that by characterizing freedom as a causality “from itself” – alluding to Kant’s definition at A446/B474 – Reinhold thinks he has protected his theory from the Free Will Dilemma. When pressed by Maimon about the determination of choice for or against the moral law, he reverts to this definition. Therein lies his problem, however, for Kant’s point is that there is only one way to think of freedom as a power beginning from itself, and that is in terms of the moral law. Answering Reinhold’s question, Maimon sallies back: “Can I not think of such a capacity? Oh yes! It is precisely how I think of nothing [Nichts],” to which he adds: “Nothing has no ground. Nothing has no results. It is not actual, not possible, etc.” (Reference Maimon1793, 240). But Maimon does not need to press these further claims. All he needs to say is that liberty of indifference amounts to “nothing” – a concept without reality – because as soon as we formulate the law of free causality we end up with Kant’s doctrine: moral necessitarianism.

3.2 Giving the Law, Being Determined

This is not an option that Maimon himself pursued, nor does it appear in the review of Creuzer’s book that Fichte published in the Allgemeine Literatur-Zeitung (Reference Fichte, Creuzer and FichteFichte [1793] 1846). Instead, Fichte appears to take Reinhold’s side by exonerating him of the charge of noumenal determinism. As Fichte initially frames the issue, Creuzer’s objection is that the definition of freedom agreed upon by Reinhold and Kant – according to which one has the “capacity [Vermögen] to determine oneself in absolute self-activity to obey or disobey the moral law, and hence to act in diametrically opposing ways” – violates the “law of logical ground” (Fichte’s label for the Principle of Determinate Lawfulness) (412). Fichte’s response is that “Reinhold has thoroughly refuted this possible objection in advance,” referring to his 1792 Letters on Kantian Philosophy. And yet Fichte qualifies his praise in the next sentence: “It is this reviewer’s conviction, which he affirms with full respect for this great and original thinker, that Reinhold has neither indicated nor overcome the basis of the misunderstanding in question” (412–413). In the rest of his review, Fichte adds a few (highly compressed) observations of his own.

As we read Fichte’s review further, it becomes clear that his deference to Reinhold is, much like Creuzer’s deference to Kant, somewhat artificial. This is not evident at first, since Fichte’s initial point is that we must be careful to distinguish between (1) the “act of determining, as a free action of the intelligible I” and (2) the “determinate being, as an appearing state of the empirical I” (413). The former is what he means by “self-activity,” his preferred term for “will” (Wille), and the latter is what he means by “choice” (Willkür) in its empirical character. The will is originally “formless,” Fichte writes, because it is what gives the moral law; it is only the restriction of the will to this principle that appears in consciousness as a “fact” of being bound to that law. For this reason, Fichte argues, the will qua self-activity never appears in consciousness, but is merely a “postulate” we must assume as the ground of our determinate acts of choice: “The will never appears as determining, but always as determined” (413). All of this is consistent with Reinhold’s account.

On the basis of this distinction, however, Fichte then asks whether self-activity of the will is the cause of what we feel in experience as a capacity to do otherwise, where we appear to be at liberty to determine ourselves in the direction of reason or sensibility:

Fichte in turn wants to separate the self-activity of the will as a determining power from the appearance of this determination as choice; for his claim is that the appearance of a capacity to act for or against the law pertains only to the latter. The point is simple enough, yet it carries important implications. What Fichte finds Reinhold guilty of doing is taking a property of choice in its empirical expression for a property of the will in itself. In saying this, he anticipates a claim Kant makes four years later in his Metaphysics of Morals:

Anticipating Kant’s point, Fichte adds that the “source of this misunderstanding can be eliminated only by returning to what seems to this reviewer to be the true spirit of the Critical philosophy” (414). For Fichte, this requires going beyond the terms of the Freiheitsstreit itself.

Up until 1793, this controversy had revolved around the dilemma of reducing the will’s activity to necessity or chance, and all the main players of the controversy had defined their positions in response to this problem, as was the case with Ulrich, Schmid, Reinhold, Creuzer, and Maimon. In Fichte’s view, this dilemma is averted once we distinguish between the free causality of the will (as Wille) and its determinate appearance in a natural series (as Willkür). Why? Because worries about noumenal determinism and noumenal chance go away, Fichte maintains, once we define freedom as a form of self-activity that does not proceed from a chain of antecedent causes. The will is what gives the moral law to itself, and through this inner restriction to the norm of its own causality it thereby determines itself. Once this determination occurs, willing manifests in a lawful form, and this is when the appearance of a capacity to act otherwise appears on the scene. The problem Fichte thinks needs a solution, then, is how Wille and Willkür can agree – how an original act of self-determination can harmonize with its empirical character.

The solution Fichte goes on to propose, in the “spirit” of Kant’s philosophy, rests on a novel version of the doctrine of pre-established harmony:

As to why we must assume such harmony in the first place – the central assumption of the argument here – Fichte offers no direct support, although he does leave the reader with a hint by writing that one must assume such harmony for the sake of a “moral world order” (moralischen Weltordnung; 414). Fichte also suggests that this need for a higher law unifying freedom and nature is “Kant’s true opinion,” and that Kant is committed to demonstrating “that freedom must have a causality in the world of the senses” (415). Among the places in his writings where such a higher law appears, the most compelling, on Fichte’s view, is Kant’s appeal in the Religion to a “higher assistance” coming to aid a human being’s moral regeneration to the good. Concluding the review, he adds that this doctrine of higher assistance is “so intimately woven” into the spirit of Kant’s philosophy that it is “laughable” that Creuzer failed to understand it (a shortcoming we may forgive Creuzer now, as this same doctrine has puzzled Kant’s readers ever since).

When Fichte wrote this review, he had yet to discover the principle that would allow him to do what Kant perhaps thought was impossible, namely, to demonstrate a causality of freedom in nature according to a higher legislation (what Fichte would later call the principle of the “pure I”). Nevertheless, this early review is valuable for showing us how the Freiheitsstreit was beginning a fresh turn in the year 1793. While Fichte finds Reinhold guilty of defining an intelligible faculty (“freedom of will”) according to its empirical appearance (“liberty of indifference”), he seems to agree with Creuzer that liberty of indifference is not a power attributable to transcendental freedom as such. We also find evidence that Fichte sides to some extent with Kant’s moral necessitarianism, as when he remarks that the pure will can determine itself in only one way, under the moral law. Still, the argument of his review shows signs of moving in a different direction, as Fichte does not think that the dilemma between determinism and indeterminism captures the central problem of freedom. As he states, the central problem is how the determinate appearance of choice in a natural series can harmonize with its freely legislated ground; and Fichte believes that only a doctrine of predetermined harmony can remove the appearance of this tension.

One cannot help but think that in saying this Fichte was sidestepping the Free Will Dilemma instead of resolving it. The problem of how my actions in a natural series of time determinations can agree with the spontaneity of my will is akin to what Kant had set up in the first Critique as the antinomy between freedom and mechanism. As we have just seen, Fichte’s proposal is that showing how a causality of freedom in nature is possible requires a doctrine of predetermined harmony, such that freedom and nature unite under a higher legislation (the principle of the “pure I”Footnote ⁵³). But even if we grant that Kant’s system needs such a higher principle, the question to put back to Fichte is how freedom of will is compatible with immoral choices, and this brings us back to the Problem of Imputation. For if my freedom of will is causally governed by the moral law, then the question of how my violations of that law count as free choices remains a puzzle even within a system of predetermined harmony. Nothing Fichte says in his review of Creuzer sheds light on this.

3.3 Kantian Indifference: Three Recent Accounts

Turning now to the contemporary literature, some two hundred years after the Freiheitsstreit, it is curious to find a number of scholars defending a version of Reinhold’s theory of indifference – with or without reference to Reinhold – on the grounds that it is the only way to explain how immoral action can still count as free action.

Reference 51Guyer and WatkinsPaul Guyer (2017), for instance, has argued that Kant is committed to liberty of indifference, one that defines freedom as the “ability to choose between alternatives,” adding that a non-liberty theory is “impossible to reconcile with Kant’s account of imputation in the Religion” (124 n). He builds his case on Kant’s distinction between Wille and Willkür along with the principle that ought implies can. Guyer suggests that it is not difficult to consider Wille to be the source of the moral law’s unconditional bindingness, and that what it discloses to the agent is its freedom of Willkür, understood as a multidirectional power to form maxims either in conformity with the law or against it. “What the moral law discloses to us,” he writes, “is just that we always can choose to act in accordance with the moral law – that we have the capacity to choose for it, even though we also have the capacity to choose against it” (137, emphasis added). The point of Kant’s distinction in the Religion, Guyer maintains, is to make this more precise by showing that one’s power of choice is always “free to choose to make the moral law its fundamental maxim or not, and imputable for choice either way” (123–124).

In drawing this conclusion, Guyer expresses agreement with a similar account put forward by Reference MorrissonIain Morrisson (2008), who claims that “Kant’s concept of practical freedom grounds the notion that one could always do otherwise (i.e. freedom of indifference)” (90). On Morrisson’s reading, Kant’s commitment to indifference follows from his notion of “practical freedom,” or the capacity to be independent from “pathological necessitation” (Kant’s phrase for the motivational force of one’s sensible nature). “To claim that we are necessitated by our sensibility,” Morrisson writes, “is just to deny that we have the ability to do other than that which our sensibility demands. To deny necessitation (as Kant’s notion of practical freedom does) is, therefore, to allow for a capacity to do otherwise” (90–91). For textual evidence, Morrisson highlights a point Kant reportedly made during lecture, to the effect that it is not possible to conceive of pathological necessitation in a human being because freedom consists in this, “that he can be without compulsion in the pathological sense; nor should he be compelled in that way. Even if a man is so constrained, he can nevertheless act otherwise” (V-Mo/Mron II, AA 29:617). For Morrisson, this is evidence that Kant viewed freedom in terms of a leeway power.

Lastly, Reference KohlMarcus Kohl (2015a) has worked to reconstruct Kant’s commitment to indifference along the lines of Kant’s distinction between perfect, or “holy” wills and imperfect, or “human” wills. What is demanded by the moral law, Kohl explains, presents holy wills with no set of alternate possibilities, because they are immediately moved to do the right thing. For human wills who are afflicted by countervailing interests and inclinations, there is no direct connection between (1) recognition of duty and (2) motivation to act from duty. As Kohl observes, “an imperfect will’s recognition of the right reasons cannot inevitably lead to right action because such a will faces alternatives or obstacles to right action”; hence, “when an imperfectly rational deliberator recognizes the force of normative reasons, any rational influence would leave her with the option to go against those reasons” (335). Indifference of choice – for or against the law – is a feature of our imperfect rational agency, a reflection of the fact that our recognition of duty does not come with sufficient determining grounds to form our maxims in accordance with the law.

By way of reply, let us begin by reexamining the text cited by Morrisson, which is admittedly one of the few places where Kant appears to align his theory of freedom with liberty of indifference.Footnote ⁵⁴ In the full passage from the student notes based on Kant’s 1785 lectures on ethics, he is reported to have said the following:

In isolation, the passage cited by Morrisson gives the impression that Kant entertained the idea that to be free is to enjoy the multidirectional power to act otherwise. A different picture begins to surface when we read it in its full context. All Kant is saying in this lecture is what he repeated time and again in his published work: that the human faculty of choice enjoys practical spontaneity, which he defines in terms of independence from pathological compulsion. The faculty of Willkür is always free in this sense of the term because human agents, unlike animal agents, are not compelled necessarily to act on their impulses and inclinations. Notice too that Kant’s passing remark about acting otherwise appears in this context, that is, when a human being is said to be constrained to act on the promptings of his sensible nature. Kant’s point is that even when such promptings appear overwhelming, our choice remains independent. To say that such an agent “can nevertheless act otherwise” is only to specify this negative characteristic of spontaneity (i.e., lack of pathological necessitation), not a further power of indifference.Footnote ⁵⁵

The rest of the passage quoted above is instructive, as we find Kant drawing a link between moral compulsion and freedom along the lines considered earlier. Because freedom of will is causally governed by the moral law, conformity of choice to the inner law of freedom admits of degrees. The more one’s choice aligns with the moral law (in forming maxims out of a recognition of oneself as their universal legislator) the more one is autonomous. This follows from Kant’s denial of liberty of indifference, since spontaneity of choice can realize its independence from sensibility by forming maxims that accord with the moral law. The more one’s Willkür harmonizes with the law of its own Wille, the more that faculty of choice is free – and Kant even adopts a special term for the perfection of Willkür, calling it “autocracy” (MS, AA 6:383). When we actualize a resolved commitment to morality, we bring the empirical character of choice closer to its original idea, the predisposition to the good making up each person’s noumenal self. The result is a genuine power over sensibility (autocracy) that aligns with the will’s essence of being a law to itself (autonomy) (see MS, AA 6:383).

This speaks indirectly to Guyer’s point about the moral law disclosing one’s capacity to choose freely “for or against” the law (Reference 51Guyer and Watkins2017, 137).Footnote ⁵⁶ As Guyer maintains, Kant believes that freedom would not be an item of consciousness were it not for a command to act in ways which (in principle) run contrary to one’s inclinations. The question to ask, however, is whether Kant intends to characterize the freedom we encounter via our awareness of the moral law in terms of a leeway power, and that I find less certain. Speaking generally, it would be odd for Kant to say that the moral law reveals, the moment we become conscious of it, our multidirectional power to act for or against this law. This would make for a mixed moral message, saying “You ought always to form your maxims in accordance with this principle,” but with the added proviso, “And you enjoy leeway freedom.”Footnote ⁵⁷ One might think that the unconditional bindingness of the moral law is meant to close off possibilities that run counter to its principle rather than opening them up. So it is not obvious to me how a power of indifference, or an equilibrium between options, could emerge from such a law itself.Footnote ⁵⁸

A more specific point to consider is that even if the moral law does not entail leeway freedom, it also does not preclude moral error. I gather that what motivated Reinhold’s theory of indifference – and what continues to make this position attractive among some Kantians today – is that if we lose a capacity to do otherwise, immoral actions can no longer be imputed to us.Footnote ⁵⁹ Nonetheless, we have good reasons to think that Kant is able to secure responsibility for evil without adopting leeway freedom. His guiding principle is that actuality entails possibility, and the empirical fact that evil exists in the world suffices to show that it must have been freely chosen, since otherwise it would not be moral evil at all. To make Kant’s position more precise, we can say that the one-way power of choice to live up to the principle of its own will excludes a genuine ability to act otherwise, but it does not exclude – and this is the key point – the possibility of corruption. My morally legislated will can demand that I ought to A, and directing my choice to A is the only exercise of my spontaneity that brings me closer to autocracy. I then lack the ability to stand indifferently between A and not-A, but this does not make not-A an impossible course of action.Footnote ⁶⁰

None of this entails that Kant is committed merely to the logical possibility of moral imperfection. As Reference KohlKohl (2015a) rightly points out, this would run the risk of trivializing Kant’s theory of evil. I agree with Kohl, on both interpretive and systematic grounds, that Kant is committed to representing evil as a real opposition to the moral law rooted in an inner “deed” (the propensity to reverse the order of incentives and place self-love above the moral law). However, I do not agree that this commits Kant to anything like a liberty of indifference view; in fact, I think it requires the opposite. Indifference implies an equilibrium between opposing forces, a zero point between not-A and A, whereas resistance implies a countervailing force, an active not-A to any given A. That Kant characterizes evil in terms of active resistance is a reason to reject rather than affirm leeway freedom, since the latter would occupy a strange place – what Kant calls the zero point between a positive and negative force – and it is not clear how well that sits within Kant’s general metaphysics of forces.Footnote ⁶¹ At any rate, we have seen in §2.5 that active resistance does not require a “power,” technically speaking, all the more so given Kant’s frequent suggestions that evil is a misuse of freedom.

Nevertheless, Kohl offers the important insight that acting contrary to the moral law must always be an “option” for imperfect rational agents like ourselves (Reference Kohl2015a, 353). I agree, so long as we distinguish an “ability” from an “option,” and specify the latter as a corruption rather than a power. To say that the moral law is the causal law of Wille does not rule out the possibility of Willkür’s imperfection and error; that was Reinhold’s mistaken inference. Moral imperfection is always a possibility for us because unlike the kind of holy agent described earlier, we are afflicted with tendencies to weakness, impurity, and self-conceit, and these tendencies create a real tension between how we in fact structure our choices and how we ought to be. For Kant, we enjoy no liberty of indifference and no equilibrium between opposing powers. We are always in the battle, as it were, struggling between a devil on one shoulder and an angel on the other. The devil is who we become when we put ourselves above the law; the angel is our true self, who is both the law’s master and servant. In the end, Kant understands moral life as an infinite task: the closer we approach the law of autonomy, the greater our freedom.