In his new book, Kant’s Conception of Freedom, Henry Allison provides a comprehensive study of Kant’s views on freedom. Though Allison’s interpretation of Kant on these issues is well-known and influential, the new book provides a distinctive angle by virtue of its developmental approach. Starting with Kant’s work in the 1750s and taking the reader through the late works of the 1790s, the ten chronologically ordered chapters of this book show how the problem of freedom threads through Kant’s work, emerges as a specific problem by the beginning of the critical period and becomes one of the central, unifying concerns around which the critical project develops. Together with Allison’s Kant’s Theory of Freedom (1990), the book constitutes an authoritative take on Kant’s view of freedom.

In this review, we first set out the argument of the book, proceeding chapter by chapter. We limit ourselves to select details and the occasional observation; we also identify the interpretive claims that strike us as significant departures from, or notable developments of, Allison’s earlier work. We then take a closer look at Allison’s treatment of a specific issue: why, and in what sense, Kant comes to reject the compatibilist position he held in the 1750s.

The narrative of the development of Kant’s conception of freedom, as Allison introduces it in his Preface, is organized around two ‘tracks’: a metaphysical track and a moral track, along which Kant’s understanding of freedom develops – and which themselves stand in an evolving relation to each other (pp. ix–x). In the first four chapters, which begin with a discussion of the Universal Natural History and Theory of the Heavens (1755) and end with a look at Dreams of a Spirit Seer (1766), readers are introduced to the metaphysical and moral concerns that drive Kant’s early philosophical efforts and that gradually converge to form a well-defined problem about the nature of freedom.

Chapters 1 and 2 focus on the metaphysical track. Kant’s early – and, as it turns out, lifelong – acceptance of the principio rationis determinantis (PRD), the Kantian version of the principle of sufficient reason, is singled out as an important, lasting constraint on Kant’s thinking about freedom (p. 13). The most immediate and consequential effect of Kant’s acceptance of the PRD is that it leads Kant to reject Crusius’ libertarian view of freedom and endorse instead a Leibnizian/Wolffian compatibilism (pp. 26–30).Footnote ¹ But as Allison notes, there is, from the very beginning, some tension between this commitment to compatibilism and other aspects of Kant’s early metaphysical worldview. In particular, Kant’s endorsement of physical influx, which can be traced back to 1755, is hard to reconcile with his compatibilism (pp. xii, 54–5, 82). Though Allison’s discussion of this point is somewhat cursory, the idea appears to be that someone who adopts physical influx construes the body as standing in real relations, whether to the mind or to other things, i.e. ‘external’ things. When physical influx is added to a compatibilist framework, it accordingly becomes necessary to explain how spontaneity – action issuing ‘from an inner principle’ (New Elucidation, 1: 402) – can be reconciled with real external necessitation. By contrast, the compatibilist who adopts pre-established harmony need only explain how spontaneity and ideal external necessitation can be reconciled – a much less ‘complex’ task (p. 82).

Kant’s compatibilism comes under further pressure from his reflections on morality, as chapters 3 and 4 show. This part of the story is less straightforward. In the early work of the 1760s, as Allison observes in chapter 3, it is the absence of explicit discussion of freedom that is most telling (p. 124). What Kant is overtly concerned about during this period is the nature of moral obligation. An important and lasting early insight, which Allison finds in the 1764 Prize Essay, identifies the distinction between conditional and unconditional obligation and argues that moral obligation must be unconditional (p. 86). This, in turn, leads Kant to articulate a number of important questions. Where do unconditional obligations originate? How do we recognize them? And how are we motivated to act on them? Allison usefully reminds readers that these questions take shape for Kant at a time when his general approach to morality is, in Allison’s words, ‘naturalistic’ and ‘anthropological’ (pp. 114, 124). But as Allison documents, Kant fails to find a naturalistic view that successfully captures the distinctive nature of moral obligation (pp. 96–7, 120–1) or the possibility of moral motivation (pp. 119–20).

Chapter 4 turns to a more direct and lasting influence on Kant’s thinking about freedom and morality: Rousseau. As Allison recalls for readers, Rousseau argues in Emile that freedom is a condition not only of choice and moral agency, but also of judgement and our cognitive capacities. Moreover, Rousseau suggests that we are aware of this freedom through a specific kind of self-consciousness that accompanies our activities of choice and judgement (pp. 141–3). According to Allison, the influence of these views is readily visible in Kant’s Remarks in the Observations on the Feeling of the Beautiful and Sublime (1764–5) and Dreams of a Spirit-Seer (1766) (pp. 152–4). But the kind of freedom necessary to underwrite these views, Allison emphasizes, cannot be accommodated in a compatibilist framework. What is needed is a capacity for unconditioned self-determination, for absolute spontaneity (p. 143). So Kant, who ‘adhere[s] to a Wolffian compatibilism’ and a broadly naturalist approach to moral philosophy when he first starts studying Rousseau (p. 132), emerges from this encounter having ‘abandoned his earlier naturalistic view’ (p. 154) and recognizes the need for a non-naturalist alternative to compatibilist freedom.

Allison’s focus in chapter 5 is primarily on the ‘silent decade’ that follows the Dissertation and lasts until the publication of the Critique of Pure Reason (1781). This transitional chapter consists primarily of analyses of Kant’s handwritten notes from the period. Allison warns readers that the ‘cryptic’ and provisional nature of the textual fragments under consideration makes the chapter a challenging one to work through (p. xv). This indeed turns out to be the case, but if we allow ourselves to simplify somewhat heavy-handedly, the most significant developments appear to be as follows. One key insight is that compatibilism secures only the ‘freedom of the turnspit’ and so is unable to underwrite morality (p. xiii). The freedom necessary for imputation is the capacity ‘to initiate a first beginning’, i.e. absolute spontaneity (p. 187; see also pp. 190–1). A second key insight argues that such absolute spontaneity is also necessary for cognition (pp. 207–9, 222–3). But the most important insight, it seems, is that there is a way to conceive of freedom as an absolute spontaneity without falling afoul of PRD. According to the view that Kant begins to sketch out in this transitional period, rational agents operate in a causal landscape in which any number of factors incline but do not determine them to act or judge. What is necessary for rational action and judgement is sufficient for free action and judgement, namely, that reason add a ‘complement of sufficiency’ by taking some inclination to be a reason to act or some perception to be a reason to judge (R5611, 18: 252, quoted on p. 210). All free actions and judgements accordingly have a sufficient ground, as PRD requires, but this sufficient ground includes the exercise of an absolutely spontaneous capacity of reason.Footnote ²

So by the time the critical period begins, Kant is convinced that ‘both the theoretical and the practical use of reason [require] an appeal to at least the conception, if not the reality, of freedom in the transcendental sense’ (p. 233). Chapters 6–8 show how this claim is developed in the critical texts. The first of these focuses on the Critique of Pure Reason (1781/1787), particularly the Third Antinomy and the Canon, two sections of the text that have been taken to offer conflicting accounts of practical freedom. This is a debate that Allison has weighed in on before, and he delivers a similar verdict here.Footnote ³ Allison also emphasizes an important way in which the doctrine of the first Critique remains pre-critical: lacking the concept of autonomy, Kant presents a view of moral motivation as resting on the hope of happiness commensurate with virtue. By the lights of the Groundwork, of course, such motivation is heteronomous and thus incompatible with moral action (p. 298).

The crucial but missing piece of the critical account of freedom – autonomy – is introduced, as is well known, in the Groundwork for the Metaphysics of Morals (1785), the third section of which is the focus of chapter 7. Allison’s discussion here largely recapitulates his previous, now-seminal, studies of these texts. He defends the opening arguments of Groundwork III but deems the deduction of the categorical imperative a ‘failure’ (p. 330).Footnote ⁴ What is it that vitiates the deduction? According to Allison, Kant thinks it is possible that we are rational automata: thinking beings under the illusion that we have wills (pp. 330–1). So we cannot infer from the fact that we are theoretical reasoners, with the spontaneity necessary for judging and employing ideas of theoretical reason, that we are also practical reasoners, beings with wills and the capacity for agency.

On Allison’s reading, then, the Critique of Practical Reason (1788) marks a significant advance over the Groundwork. The ill-fated attempt to give a deduction of the categorical imperative from a non-moral premise is replaced by the doctrine of the fact of reason, which Allison characterizes as the ‘consciousness of the moral law as supremely authoritative or, equivalently, the consciousness of the categorical imperative’ (p. 372). And through the fact of reason, Allison argues, the objective reality, as well as the actuality, of freedom is established. But the claim that we can establish the actuality and reality of freedom raises new questions about the relation between theoretical and practical reason; hence the task of establishing the unity of reason becomes central to the second Critique (pp. 352–3).Footnote ⁵

The last two chapters turn to the 1790s and Kant’s ‘final thoughts’ on freedom (p. 451). Chapter 9 focuses on the Critique of the Power of Judgement (1790); chapter 10 turns to the Religion within the Boundaries of Mere Reason (1793) and the Metaphysics of Morals (1797). The latter chapter defends the Wille/Willkür distinction, characterizing it as a distinction between the ‘law-governed’ and ‘voluntaristic’ elements of Kant’s theory of freedom. As Allison points out, these elements were rendered ‘inseparable’ by the Reciprocity Thesis, for if a free will is a will under the moral law, as the Reciprocity Thesis claims, then a voluntarist will is a law-governed will. As voluntarist, it has a freedom that is not the freedom of the turnspit; as law-governed, it has a freedom that is not the freedom of indifference. Such a will accordingly occupies the long-sought ‘logical space between the freedom of a turnspit and the liberty of indifference’ (p. 464); and it seems that Allison might draw the overarching narrative of his book to a satisfying close with his discussion of the Wille/Willkür distinction.

However, the rest of chapters 9 and 10 and their discussions of virtue and evil, respectively, unsettle this otherwise tidy conclusion. The propensity for radical evil, Allison argues, is the propensity to prioritize self-love over the good (p. 481). In calling this evil ‘radical’, Kant is not characterizing the ‘magnitude’ of the evil, but its ‘rootedness in human nature’ – which is what accounts for the fact that the acquisition of virtue is inevitably a struggle for us (p. 475).Footnote ⁶ In these closing chapters, Allison not only points to an intriguing connection between virtue and radical evil, he also characterizes both as manifestations of the ‘struggle of freedom with itself’ (pp. 428–9; see also pp. xvi–xvii). What this suggestive claim entails about Kant’s doctrine of freedom – indeed, whether it hints at the ultimate untenability of Kant’s theory or shows that the theory correctly captures something about the conflicted nature of human freedom – is left for readers to work out.

Before closing, we take a closer look at one of Allison’s interpretive claims. Recall that in Allison’s account of the development of Kant’s views on freedom, the pre-critical half tells the story of how and why Kant comes to reject compatibilism. In the 1750s, Kant’s view is that freedom of action is perfectly compatible with – and in fact requires – its being fully determined. As Kant puts it in the New Elucidation of 1755,

The critical break with this compatibilist view is summarized by Allison in the following way:

Though Allison does not make this totally clear, Predeterminism is the claim that, for every act A that occurs there is a (temporally) prior event E such that the occurrence of E, along with the laws of nature, necessitates A. In contrast, Determinism is the claim that for every act A there is a ground through which that act is necessitated. Predeterminism is a species of determinism, but it is not identical with it. So in the above passage, Allison points out that, in rejecting compatibilism, Kant denies that the free will is ‘predetermined’, that is, determined by a temporally antecedent ground. He does not, however, claim that the free will is not determined, in the sense that free acts of choice are not necessitated.

In our view, Allison is right to point to there being an important distinction between the two kinds of determinism and, though Allison does not make this point explicit, he is also right to treat as significant the fact that this distinction is not available to the pre-critical Kant, who considers time to be transcendentally real. However, Allison fails to explain why time plays such an important role in Kant’s argument. Kant’s formulation of time as a principle of our receptive sensibility entails that, so long as we are still operating within a compatibilist framework, we must construe all acts as sensibly conditioned, in the sense of being determined in time. Given that the pre-critical Kant, as well as the Leibnizian/Wolffian, construed free willing as depending on the (spontaneous) activity of the intellect, the fact that the agent’s activity would be at least partly based on a non-intellectual principle (i.e. time) would entail that the rational will’s acts were not in fact spontaneous, and thereby not free.Footnote ⁸ This means that Kant’s move to conceiving of time as a law of sensibility, and thus as ‘alien’ to the intellect, is a watershed moment in the development of Kant’s conception of freedom, one that is not given due attention in Allison’s narrative. This move raises a problem for Kant’s conception of freedom that can only be resolved by transcendental idealism, which makes (at least epistemically) possible a genuinely (by Kant’s lights) unconditioned form of spontaneity, in the sense that all spontaneous acts are determined entirely through the nature of the intellect, rather than through the intellect’s activity as conditioned by sensibility in time.

Allison’s narrative foregrounds the idea of a break between the pre-critical and critical Kant; but the preceding considerations suggest it is possible to construct a narrative that brings out the continuity between the pre-critical and critical Kant. When Kant rejects compatibilism, as mentioned earlier, he rejects predeterminism but not determinism. Thus the breach with compatibilism is not as complete as Allison sometimes suggests; in particular, the compatibilist idea that freedom is not only compatible with but requires determination is preserved in the critical view that an act is free just in case it is determined by a spontaneous intellectual capacity, through the nature of the intellect (and reason in particular). According to this alternative developmental narrative, Kant consistently holds a form of ‘source compatibilism’ throughout his career, and the important development of Kant’s view occurs in his coming to see that this source compatibilism cannot consistently be maintained by the transcendental realist about time.Footnote ⁹

We have suggested that an important moment in the development of Kant’s theory of freedom is overlooked in Allison’s account – and that the inclusion of this moment allows the narrative arc as a whole to take on a different shape. If this suggests that the specifics of Allison’s developmental account are debatable, it does so by following Allison’s lead and looking for an interpretation that takes in the full range of a philosopher’s evolving concerns and commitments.