In The Problem of Free Will and Naturalism: Paradoxes and Kantian Solutions Christian Onof proposes an interpretation of Kant’s theory of freedom aimed at offering an alternative to naturalistic theories of free will in the current analytic debate. Onof’s book does not force Kant’s texts into providing answers to present-day questions but instead provides a comparison between some Analytic theories of free will and what the author takes to be Kant’s libertarian compatibilism. The book will be of great interest to Kant scholars as Onof proposes a novel libertarian interpretation of Kant that does not rely on a two-world metaphysical framework as has traditionally been the case, but rather on a one-world two-aspects framework.

That being said, the book is written in language accessible to both analytic and other Kantian researchers. For the analytic reader, all of the current debates the book addresses will be common knowledge, while the Kantian exegesis is written in an accessible manner, although some familiarity with Kant’s practical works would be beneficial (especially the Critique of Practical Reason and the first section of the Religion within the Boundaries of Mere Reason). Likewise, readers will find the opening chapters on present-day debates regarding compatibilism and libertarianism quite clearly explained, although it is worth mentioning that Onof does not offer an overview of the entire analytic free will debate, but only of issues directly relevant to his interpretation of Kant.

The book’s argumentative strategy is as follows: because of their naturalist framework, analytic compatibilists and incompatibilists fail to argue that we are agents endowed with free will. On the compatibilist side, Onof discusses the variants of source compatibilism, namely, real-self or mesh theories, and Fischer and Ravizza’s reasons-responsive account. The former fails because it cannot explain why we are free when we act against values and norms adopted for ourselves. Fischer and Ravizza’s account is meant as a solution to this issue. However, Onof argues that their account is exposed to Pereboom’s argument from manipulation, according to which we are not free agents insofar as we cannot own the reasons-mechanism that leads to actions. The incompatibilist side, through event-causal (Robert Kane) and agent-causal libertarianism (O’Connor), might provide encouraging answers to the manipulation argument. However, they fall prey to what Onof calls the Rational Luck Objection. While the traditional luck objection concerned the randomness of occurring events (including choices) and targeted figures such as Christian Crusius, the Rational Luck Objection concerns the agent not having further grounds for choosing the reasons for her actions, thus arriving at an explanatory impasse (and implicitly at randomness in choosing these reasons for action). The rest of the book is a development of a Kantian account of freedom meant to solve these very issues.

The book’s sequence of chapters closely follows this argumentative strategy. The first four chapters consider the analytic theories mentioned and an initial sketch of some Kantian answers to these issues. Thus, Chapter 1 presents the debate surrounding compatibilism and Chapter 3 the debate surrounding libertarianism (in its event-causal and agent-causal incompatibilist forms). Chapters 2 and 4 present Kant’s metaphysical account of freedom in terms of how it can solve issues from the analytic debate: Chapter 2 discusses the compatibilist issues from Chapter 1, while Chapter 4 discusses the libertarian issues debated in Chapter 3. Chapter 4 is especially interesting as it provides a novel interpretation of the Third Antinomy’s Resolution chapter, arguing that it is not merely logical possibility that Kant seeks for transcendental freedom but real possibility as a self-standing causal power with effects within nature.

Chapters 5 to 7 fill in the details of Onof’s interpretation of Kant, which consists of a metaphysical, volitional, and psychological account of freedom (more on each below). Chapter 5 presents the volitional account of freedom, while Chapter 6 presents the psychological account. Chapter 7 presents the unitary picture that emerges from the relation among all three accounts and how they answer the challenges raised by the debate regarding compatibilism and incompatibilism.

Onof’s argument in Chapters 5 to 7 is that Kant’s transcendental idealism can provide answers to the mentioned issues faced by compatibilists and incompatibilists, respectively. He does so by separating Kant’s theory into three accounts of freedom: a metaphysical account, a volitional account, and a psychological one. Since Onof proposes a one-world but two-aspects interpretation of Kant, the foundational world is that of things in themselves with its two aspects being the intelligible and the empirical character of action. In correspondence to this framework, the metaphysical account will ground the volitional and psychological accounts. Let us now see what each account consists of:

(1) The psychological account of freedom is concerned solely with the empirical character of action. This account is in need of a ground because it concerns only the agent’s temporal mental states, drives, temperament, and other such empirical causal factors, and given that determinism is true at the empirical level of action, they can only be grounded in the intelligible character of action. They are hence only sensible schemas of the choices we make. So, a maxim of action occurs as a temporal mental state (on the side of the empirical character of action), but the pursued goal is chosen through my power of choice (Willkür) and through a maxim prescribed by my practical reason (Wille).

(2) The volitional account will ground this psychological account insofar as my power of choice and practical reason are part of what Onof calls the volitional account of freedom and are part of the intelligible character of action. The volitional account thus concerns our practical freedom with everything that has to do with how we make our choices and also the assumptions we make about our transcendental freedom. Onof takes Kant to propose through his conception of practical freedom a volitional account lying in the agent’s own conception about her will and how she makes her own choices. This includes a conception of Willkür (through which the agent chooses what goal to pursue) and Wille (through which the appropriate maxims for the chosen goals are prescribed), but also of freedom as a regulative idea of acting independently of all natural conditions as a condition for an agent’s choice to be her choice. The latter is the case because the law of the intelligible character is the moral law, and to act morally, the agent has to be able to act independently of all natural conditions. However, the multitude of maxims is in need of a unity, and this is supplied by transcendental freedom in the metaphysical account.

(3) The metaphysical account does the work for Onof of grounding the volitional account (and implicitly the psychological one) because it concerns transcendental freedom. While the grounding relation between the metaphysical and psychological account is straightforwardly causal, the grounding relation between the metaphysical and volitional account is teleological. The law of the agent’s intelligible character is the a priori and pure moral law, which is prescriptive and not descriptive. Namely, the agent chooses the fundamental disposition, orientation, or value worth pursuing in her life, one’s Gesinnung. The moral law requires she adopt the moral Gesinnung, but she can choose otherwise at any stage of her life. When choosing the moral Gesinnung, this acts not as something necessitating actions at the empirical level, but as offering a moral ideal that the agent tries to attain with each and every action. Of course, the freedom the agent enjoys in the metaphysical account is outside of time and independent of natural laws and causality while determinism is still true at the level of appearances. This is why Onof labels his interpretation as a form of ‘sourcehood’ compatibilism. Further, because of the leeway-freedom the agent enjoys in the metaphysical account, this account is also libertarian (though normative).

In terms of the issues Onof has set out to solve, this interpretation of Kant is relevant given that our freedom is outside of time and therefore it is not subject to manipulation arguments. And given the normative dimension of this libertarian interpretation, it is not subject to the Rational Luck Objection either: the agent is required to choose (and develop) a moral Gesinnung, but has the leeway to act against it. Asking why an agent chooses to act for or against a moral Gesinnung cannot be further answered since there is no other explanatory level outside the agent’s intelligible character. Therefore, Onof reminds us of Kant’s explanations in the Religion that we cannot explain why the agent chooses to adopt either a good or an evil fundamental maxim for her actions.

An analytic reader might be left to wonder why such a Kantian account of freedom labels itself as both ‘libertarian’ and ‘compatibilist’, given that these terms together form an outright contradiction. The issue is of course discussed by Onof in several chapters of the book, and he does offer an overview of how Kantian scholarship has treated the issue. The most important aspect to keep in mind, however, is that analytic libertarians still assume naturalism. For example, for Robert Kane, the indeterminism of self-forming actions (SFAs) is an indeterminism within nature, hence his need to discuss concrete neural examples in which indeterministic SFAs do plausibly occur. However, for Kantian libertarians who are also compatibilists this ‘indeterminism’ does not occur at all in nature, but at the level of things in themselves. (That is not to say that there are no Kantian libertarians who are also incompatibilists and who may argue that practical freedom – which is empirical – does display some modicum of indeterminism.)

With that said, I would like to stress what I think are some key innovations Onof brings to the table when comparing his libertarian interpretation to traditional libertarian interpretations of Kant’s theory of freedom (Wood Reference Wood1984 and Watkins Reference Watkins2005) and what critical discussion they might generate.

Traditional libertarians assume a two-world interpretation of freedom such that there are two types of properties and causalities inherent in the agent. At the intelligible level, the agent chooses, from several possible inner natures (or characters), the one which will determine her empirical character, which in turn necessitates that agent’s actions. This is problematic in at least three ways: firstly, there is equal weight given to each nature or character the agent might choose for herself, despite Kant’s contention that the moral law is meant to prove the agent’s freedom; secondly, there is no leeway at all for the agent at the empirical level since all actions are necessitated by the empirical character (itself determined by the intelligible character); and finally, practical freedom is itself merged with this empirical character of action.

How does Onof’s interpretation deal with these problems? The first issue is solved insofar as the moral law is a normative law at the intelligible level, and thus freedom is defined through morality, while the agent still has the ability to choose an evil Gesinnung. For recent moral necessitarians (e.g. Ware Reference Ware2023) this might still prove problematic since at the intelligible level the agent can only act morally, hence the moral law would be descriptive rather than just prescriptive.

The second issue is the one Onof focuses on the most insofar as the moral Gesinnung is only a moral ideal, hence can prescribe several maxims the agent might choose (in the volitional account), but never necessitates either the agent’s empirical character or any of her particular actions. In other words, the future is open-ended, and there is temporal leeway in that regard. The examples Onof offers concern the repetition of certain outer circumstances. While I may today lend some money to a person when asked, based on further knowledge and evaluation of this situation, if the same person asks me for money again tomorrow, I might refuse based on how I interpret the actions from yesterday. While this is enough for providing at least some temporal leeway at the empirical level and as such does not allow for predictions of actions, I think Kantian incompatibilists might still argue that this is not enough for the agent to have leeway as far as present deliberations and decisions are concerned. This is because at this present moment, the agent is at some stage of development of her Gesinnung; thus, her current maxims and values necessitate the decision she makes at the time. This, of course, would fall in line with traditional objections raised by Kantian incompatibilists against compatibilist interpretations of Kant.

Finally, regarding the third issue, I find the greatest merit in Onof’s account to be that of offering an interpretation of practical freedom that is independent (as a volitional account) from a psychological account of freedom. This proves important for future research insofar as practical freedom can be studied by itself and without being wholly assimilated under the empirical character of action. Onof largely follows Allison’s own account of practical freedom here (Reference Allison1990), but with some key differences. In Kant’s mid-1770s L1 (Pölitz) metaphysics lectures, Allison argued that transcendental freedom was exclusively a cosmological causal power that is independent of all natural causal factors, while human agents are affected by natural factors, and as such only have a weaker libertarian sort of transcendental freedom, namely practical freedom. Thus, transcendental freedom was merely a regulative idea that on closer inspection might be a natural causal power after all. For Onof, however, practical freedom is not a type of transcendental freedom. Instead, transcendental and practical freedom designate the same freedom, with the former being a causal power we do possess (hence Onof’s claim in Chapter 4 that Kant needs to argue for the real possibility of transcendental freedom). And as already mentioned, practical freedom contains our conception of our acting independently of all natural conditions.

What I find problematic with Onof’s account of practical freedom is that in conceiving of ourselves as independent of all natural conditions, we are left with a contradiction with Kant’s claim that practical freedom is merely an independence from the necessitation by stimuli. As Onof argues (p. 104), the agent’s practical freedom means that the agent (1) ‘would, in her practice, understand herself as able to do otherwise’, which in turn implies that (2) ‘she conceive herself as able to act independently of natural causes’, with (2) assuming independence of all natural causes such that (2) entails that the agent (3) ‘conceives of herself as endowed with the power to “begin a series of occurrences entirely from itself” (A534/B562)’. I do not agree with (1) implying (2). If this were the case, then practical freedom requires that the agent conceives of herself as acting independently of all natural causes (much in line with the cosmological notion of transcendental freedom). This clearly contradicts Kant’s definition of practical freedom at A534/B562 as merely ‘the independence of the power of choice from necessitation by impulses of sensibility’, and the further claim that our power of choice is only ‘an arbitrium sensitivum’. Practical freedom does not seem to require for Kant that the agent conceive of herself as independent from all natural conditions. Additionally, in his Review of Schulz (1783), Kant says that the fatalist also is practically free and acts ‘as if he were free’. It would be contradictory to argue that the fatalist believes at the same time that (1) he does not act independently of all natural factors (as he would when denying transcendental freedom), but through his practical freedom and (2) conceive of himself as acting independently of all natural causes: ‘the most confirmed fatalist, who is a fatalist as long as he gives himself up to mere speculation, must still, as soon as he has to do with wisdom and duty, always act as if he were free, and this idea also actually produces the deed that accords with it and can alone produce it’ (8: 13). A way out of this may be to argue that practical freedom concerns not the first-person perspective (the agent’s conception of herself), but the second-person perspective, i.e., our social practices for assigning blame and praise. Onof, however, argues against this by saying that such a move would lead to practical freedom, assuming that we only possess a fictional freedom, i.e., a fictional independence from natural causes, much in line with Peter Strawson’s own compatibilist view (pp. 101, 105). If Onof were to argue like this, then I think he would, to some extent, also have to revise the claim that Kant argues for the real possibility of transcendental freedom in the Resolution chapter.

We often find Kant’s philosophy interpreted through the lens of contemporary debate, but with Onof’s book, we find analytic metaphysics scrutinized through a distinctively Kantian lens. Given the originality and insight Onof has provided here, there is no libertarian interpretation of Kant that can overlook this book in the future.