ERIC WATKINS (ed.) | Kant on Persons and Agency | Cambridge University Press 2018


By Christian Onof 

This volume of papers features an excellent line-up of many among the most influential contemporary Kant scholars so that the reader is entitled to expect much food for thought. And this expectation is fully met with a selection of thought-provoking papers around the topics of the person and agency, therefore dealing both with Kant’s theoretical and practical philosophy. The types of paper range from careful analyses of Kantian texts (from the Critical period) to developments of ideas that have a Kantian origin, but sometimes move well beyond that. This mixture is welcome: Kant’s philosophy is kept alive through new interpretations of the letter of his philosophy and through drawing upon insights taken from his writings and developing them in directions that may be said to be in the spirit of Kant’s writings but clearly go beyond its letter. Of course, this raises the question of how far beyond the letter one can go while remaining within the spirit of the Critical Kant, as we shall see.

1. The Unconditional Good and Our Duties

Eric Watkins starts his paper by drawing our attention to different senses of ‘unconditioned’ in Kant’s practical philosophy (pp. 12–13). This is perhaps first and foremost found in the notion of an unconditionally good will in the Groundwork (GMS). Watkins argues that this is a sense that is analogous to that found in his theoretical philosophy, and which is also found in the unconditioned as the totality of conditions of the object of pure practical reason, which Kant identifies as the highest good in the Critique of Practical Reason (CPrR, AA 5:108). This sense is distinct from the sense in which the Categorical Imperative (CI) is unconditional, since this is just the claim that its normativity is independent of any condition (GMS, AA 4:420). The unconditionality of lawgiving that characterises autonomy (GMS,  AA 4:436) is a third sense that does not sit well with the claim that the good will is the only unconditional good.

Watkins narrows down his investigation of unconditionality in Kant’s practical philosophy to three questions of which I examine the two main ones here. He first investigates the unconditional goodness of the good will. A detailed analysis leads him to distinguish the unconditioned/conditioned distinction from the intrinsic/extrinsic one (pp. 16–17). Rae Langton (2007) had already pointed to the fact that something can confer value on itself, which would mean that while it is intrinsically good, it is conditionally so. Watkins adds that, conversely, it is not clear that unconditionality must entail intrinsicality.

Watkins then suggests as an alternative approach, that a condition be understood as ‘a certain kind of metaphysical dependence relation that also functions as an explanatory principle’ (p. 17). Armed with this interpretation, the unconditionality of the good will can then be seen to amount to its not depending on anything else for its goodness, and there not being any further explanation available as to why it is good (p. 18). Watkins carefully analyses Kant’s claims about the good will in the first section of the Groundwork. This is where he points to a problem in the constructivist (e.g. Korsgaard’s) view that the good will is good insofar as it confers goodness upon itself. While Langton argues that this would entail that the good will would be conditional (upon this conferring), Watkins chooses another tack: insofar as conditioning is an ‘asymmetrical metaphysical dependence relation’ (p. 22) like causality, it follows that nothing can be its own condition since Kant consistently rejected the idea of anything causing itself. While it is open to the constructivist to reject Watkins’s metaphysical interpretation of conditioning, she would then have to provide another account of the notion of conditioning that is plausible for Kant’s uses of this term across his practical and his theoretical philosophy.

This interpretation enables Watkins to support Kant’s claim that there is at least an entailment from ‘intrinsic’ to ‘unconditioned’ (p. 21), and to address other puzzles raised by the notion of unconditionality in Kant’s practical philosophy. The unconditionality of the CI can now be seen as pertaining not to its value (as with the good will) but to its necessity (p. 23). And its normative force is unavoidable since, unlike hypothetical imperatives, it does not depend on any condition such as a commitment to satisfying some desire which we could elect to give up (p. 24).

The second question Watkins examines is that of Kant’s also assigning unconditional goodness to universal legislation. Drawing upon Karl Ameriks’s understanding of the good will, he argues that universal legislation makes the good will possible, in the sense of its having a relation to the moral law: it is thus not itself unconditionally good, but a condition of the unconditional goodness of the good will (p. 25). Kant’s claim that autonomy, and therefore universal legislation, is ‘the ground of the dignity of human nature and every rational nature’ (GMS, AA 4:436), can indeed be seen as supporting Watkins’s proposal (p. 26). Overall, Watkins’s proposal to understand conditions as explanatory dependence relations, although its metaphysical content may not seem immediately attractive to some Kant scholars, does have the merit of shedding light on a set of puzzles pertaining to interrelated concepts around the notion of unconditional goodness of the good will.

Allen Wood examines the vexed question of the extent to which the Formula of Universal Law (FUL) or the variant form of the Formula of Law of Nature (FLN) define a test of permissibility of an agent’s maxim of action. As Wood indicates (p. 31), there is a whole body of secondary literature devoted to showing in what way these formulas are inadequate, or are in fact adequate once they are interpreted properly, or can be reconstructed in line with the spirit of Kant’s original versions so as to define adequate tests of permissibility. Wood’s paper does not engage directly with this discussion, but seeks rather to show that the formulas are not intended to function in this way. That is, one does not decide what to do by submitting one’s maxim to a FUL/FLN test.

Wood draws directly upon Kant’s unambiguous characterisation of FUL and FLN as a ‘standard’ or ‘canon of moral judgment’ (e.g. GMS, AA 4:402, 403–4, 424). As such, they cannot be applied directly to yield a moral judgement. Wood draws upon Kant’s theory of determining judgements: these are used to apply ‘general concepts or rules to particular cases’ (p. 32). They themselves cannot be rule governed otherwise a regress would ensue. The faculty of judgement can, however, be trained, and Wood singles out the Doctrine of Virtue as an example of a text where Kant provides examples of particular cases to illustrate how the duties he is examining are applied (MS, AA 6:423–4, 426, 428, 431, etc.).

If FUL and FLN are to serve as a canon of moral judgement we should, according to Wood, interpret them as providing an aid to our faculty of practical judgement in cases where it may be led astray (p. 33). And Wood makes a good case for the claim that the examples Kant chooses are meant to play this role of guide to our moral judgement. This he does first by drawing upon Kant’s own remark that there is a pattern to violations of duty which involves our ‘taking the liberty of making an exception for ourselves’ (GMS, AA 4:424) while accepting the moral law as principle of morality. That suggests that it is indeed a clear case of our judgement getting it wrong in moving from the principle to its application. And second, he reminds us of what Kant says in the first section of the Groundwork, namely that the common human understanding is as, if not more likely to get it right when it comes to deciding how to act morally or how to judge some action (GMS, AA 4:405).

While this amounts to a pretty strong case for not viewing the FUL and FLN as tests to be used to determine whether some particular maxim of action is morally permissible, a worry arises. This is namely that, even if we accept that, in practice, we do not need to apply FUL/FLN to determine whether a particular maxim of action is permissible or not, the role of FUL/FLN is unclear. How can a test help us determine how to apply a general principle to a particular case without amounting to a rule doing precisely what Kant seems to be excluding on pain of regress, namely governing the practice of (practical) judgement? Moreover, Kant himself claims that the duties which these tests identify are derived from the FUL (GMS, AA 4:421). Wood claims that what Kant means by that is that a derivation of the duties is possible from another formula of the categorical imperative (the Formula of Humanity). This is not very convincing as an interpretation of that sentence, although by doing that, he is trying to address the worry that we do, after all, have a rule telling us how to apply a general principle.

I think that Wood’s solution is a possible one, and his paper offers more interesting reflections on how Kant tackles the individual examples. I would suggest that another way to address this worry while taking on board Wood’s important point about these formulas as a guide to moral judgement, is to argue that the FUL/FLN does not tell us what to do, but merely delimits the domain of what is permissible. In so doing, it leaves more than one course of action available, and the choice between the different options will not be morally neutral, i.e. there will be different duties competing for my attention. This does not mean that these duties are necessarily in conflict (Kant does not accept real conflicts of duty), but that judgement will be required. If, when I am pursuing the task of developing certain talents I have and am constantly interrupted by the need to help other agents, I am going to have to exercise judgement to know how to negotiate these competing demands on my time.

And in support of my proposal I would argue that, much as the maxims that are tested are specific, they are just instantiations of what are still fairly general policies which define duties. So, for instance, with the maxim of promising, what is at stake is acting in a way that I will benefit from the practice of promising while not abiding by its rules. In that sense, while I agree with Wood’s point that ‘the maxims in Kant’s examples are never offered merely as general policies or rules of life’ (p. 37) I would thereby emphasise that the use of ‘merely’ does not, and indeed should not exclude that they do thereby define general policies about which the issue of how to apply them in particular cases will arise. When Kant describes these duties as derived from the FUL, this does not mean an analytical derivation of course, but simply that, for a given practical problem, e.g. how to behave with respect to the practice of promising, one can derive a rule by using the FUL.

Wood’s paper concludes with a section devoted to showing that the FUL/FLN cannot in fact be used as tests of moral permissibility because of the well-known problems of false negatives and false positives. He offers an interesting analysis of cases of false positives and false negatives. On the issue of false negatives, the way Korsgaard (1996) for instance analyses the application of the FUL is helpful in dealing with such cases. As for false negatives, I take it that the problem lies with the excessive specificity of the maxim, which is precisely what my proposal is designed to avoid. So, let’s take Wood’s example of a maxim of false promising only on a certain day of the week, to a person with certain characteristics etc. As Wood argues, this kind of maxim will not lead to any contradiction (however one interprets this notion) if universalised because it may for instance be the case that the ‘present false promise is the only one that would ever be made following this maxim’ (p. 41). Wood rejects the claim that this is not the maxim that the agent is acting on and, further, dismisses the strategy of wanting to have all the required information about ‘the situation and the agent’s intentions’ (ibid.).

I would argue that the first claim can be defended by making a move that is exactly the opposite of that which seeks greater specificity. If, as I suggested above, we take what the FUL has established to be that it is not permissible to seek to benefit from the practice of promising without following its rule, then we can see that the agent is acting on this maxim (now understood as a subjective principle that is more general). And we do not need to know more about the agent’s intention: insofar as she is rational, she is making this false promise on this particular day… because she stands to benefit by it (and therefore also to benefit from the existence of the practice of promising), so she is clearly acting on the more general maxim. Why should one consider this more general maxim? Because it is important that the (non-)universalisability should not result from empirical contingencies that are independent of the nature of the proposed action. For instance, it may be that certain forms of discrete promise breaking do not lead to the practice of promising being compromised. All that this indicates is that we have a sophisticated free-loader who not only wants to benefit from the practice of promising while breaking its rules, but additionally do it in such a way that it is not noticed.

There is of course much more to be said here, but I think that the most important message of Wood’s very thought-provoking paper is that the FUL/FLN test cannot tell us what to do: there is a key role for non-rule-governed moral judgement.

Barbara Herman tackles an issue that has grown in prominence over the past decade, namely that of the place of animals in Kant’s ethics. From the outset, Herman talks of bringing animals into Kantian morality, thereby suggesting that this is a paper that goes beyond the letter of Kant’s ethics. But she simultaneously announces two ‘desiderata’, the second of which is that any ‘renovation required […] not undermine the basis features of the Kantian system’ (p. 174). And indeed, Herman is very careful throughout the paper to ensure that this desideratum is met.

Herman starts by giving a broad outline of where one might bring animals into Kant’s system, and she locates this as the need to be concerned with the way we change the world when we create our moral habitat. As she remarks in a footnote, this can also be the way in which we find a place for an ethical concern for non-sentient entities, but she wisely chooses the focus on the issue of animals in this paper, for this problem is difficult enough as it is and, as she says ‘[t]he space of answers to the animal question is not cluttered with great successes’ (p. 176).

Herman rehearses the received view of Kant’s ethics which has it that only those beings that are ends in themselves, and thereby moral agents, can be moral subjects. She reconstructs a derivation for this claim that hinges on the idea that morality requires us to reason well (in line with the normative role of the Categorical Imperative), and that precisely involves us representing ‘all other moral agents equally as co-reasoners and so as moral subjects’ (p. 178). She then makes the important move of inferring that ‘animals could enter the Kantian account if and only if we are compelled by principle(s) of good reasoning to regard how they fare as an element or a condition of (our) good reasoning’ (ibid.).

This is the line adopted by Wood and Korsgaard. Wood draws on the idea that if animal behaviour that is sufficiently similar to ours is disrespected, this will be inimical to our respecting rational nature. Herman rightly points out that this would not ensure that we treat animals with ‘non-lovable behaviours’ (p. 179) ethically. Korsgaard draws on our being (partly) animals to argue that what is good for us includes what is good for us qua animals, so that what is good for animals already features in Kant’s ethics on her reading. As Herman points out however, this involves viewing morality as dealing with managing goods that are external to it, while she, I think rightly, and in line with Kant’s outlook, views it as morality’s task to determine what is good (p. 180).

The rest of the paper is devoted to a careful analysis of a passage in the Amphiboly section of the Metaphysics of Morals (MS, AA 6:443) in which Kant talks about duties with respect to animals, but in a way which seems to suggest that it is only because of a deleterious effect upon the ‘affective system that supports us in our duties to human beings’ (p. 181). This would not amount to a satisfactory account, both because it relies on some empirical fact about how our natural dispositions would change, which may not be true of certain moral agents for instance—would they thereby be excluded from such a moral requirement; and what about our attitude to animals such as cockroaches which are not likely to affect these dispositions? (p. 182).

By a thoughtful analysis of the way animal behaviour plays a role in our lives in that we understand ‘ourselves as embodied in and through encounters with a range of animal life forms’ (p. 184), Herman develops a convincing alternative reading of the Amphiboly text such that shows how mistreatment of animals would impact our ability to carry out our duties. To do this, Herman briefly surveys the other two duties that Kant identifies explicitly in the Amphiboly, namely (i) a duty with regard to the beautiful insofar as the disposition to love something without any interest in morality ‘prepares the way’ (MS, AA 6:443) for morality; (ii) a duty with regard to God that promotes the idea of the purposefulness of the universe and thereby promotes the moral incentive. All three duties ‘shape our responsibility’ (p. 190) as moral agents. Towards the end of this thoughtful, clearly written and thought-provoking paper, Herman concludes that ‘[t]his does not give us a simple story about our moral relations to animals, but there is no reason why we should expect that’ (ibid.).

2. Freedom and Autonomy

Marcus Willaschek examines two problems about the status of the proposition that we are transcendentally free as it emerges from the claims made in the Critique of Practical Reason. The first (Problem I) is that of explaining why Kant offers no argument for the postulate of freedom in the Dialectic, unlike the treatment he gives of the postulates of immortality and of God. Second, Problem II is a tension that requires resolution, namely that between freedom as an object of knowledge (because of the fact of reason) and the object of one of the Postulates of Pure Practical Reason, and thereby an object of belief.

Willaschek examines the interpretative moves made in the secondary literature and shows convincingly that none of these is entirely satisfactory (pp. 104–6). These either involve revising Kant’s claim that freedom is a postulate, either by suggesting it is a different notion of freedom that is at stake (Beck) or that in the case of freedom, our knowledge of freedom is also describable in terms of a postulate (Klemme). While these moves have obvious immediate problems (e.g. for the latter, if I know something, it is not a mere object of a postulate which involves mere belief), they make Problem I more pressing. Other interpreters (von Platz) solve Problem I by claiming that the postulate of freedom is proven in the Analytic, but how could it then be a postulate as the Dialectic of the Critique of Practical Reason defines it?

Willaschek carefully examines the proof Kant offers in the Analytic, where Kant derives the claim that the ‘human will is (transcendentally) free’ (p. 106) from the fact of reason together with the so-called ‘reciprocity thesis’ as Henry Allison named it. The latter is the claim that there is a reciprocity between being transcendentally free and being subject to the moral law. Willaschek refers to this derivation as the factum argument. What his careful analysis draws attention to is the need to tread carefully in understanding the status of the fact of reason, namely the claim that we are conscious of being subject to the moral law. He opts for interpreting it as a theoretical sentence (i.e. one about what is the case) as opposed to a practical one (i.e. one about what one ought to do). This choice is questionable, but the success of his interpretation does not hinge upon it, as long as one accepts that this fact of reason is a proposition that has some truth value. What then is our attitude to this truth-value? Some (Patrick Kain) have claimed that it is practical knowledge, and indeed, this might be considered as the default interpretation insofar as its being theoretical knowledge is excluded: we have no theoretical warrant as Kant says repeatedly (p. 108).

Here, Willaschek makes an important observation, namely that Kant does not use the expression ‘practical knowledge’ in his Critical works. Willaschek then draws upon Kant’s definition of practical cognition in the Jäsche Logik, but extends it in what appears to be very much in the spirit of Kant’s definition: Kant claims that we have practical cognition of either imperatives, or, the grounds thereof. Since the moral law is the ratio cognoscendi of freedom, the latter is grounded in it, rather than a ground of it. This is where Willaschek proposes to extend Kant’s definition of practical cognition to include what is grounded in imperatives. The upshot is that the factum argument establishes that we have practical cognition of our freedom, not knowledge (pp. 105–9). This distinction enables Willaschek to address Problem II on the condition that this claim of cognition, unlike that of knowledge, is compatible with the claim of mere belief of the postulates.

Problem I is then resolved by showing that this factum argument is what establishes the postulate of freedom. Indeed, this argument establishes the practical necessity of the belief that we are free, while the theoretical undecidability of this claim for Kant is clear from several of his statements to that effect. And since practical necessity and theoretical undecidability are the hallmarks of a postulate of practical reason, we apparently have an answer to Problem I (pp. 110–11).

It is worth adding that, among the textual evidence Willaschek can adduce in support of his interpretation, the remark Kant makes in the Critical Elucidation (CPrR, AA 5:94) actually spells out the claim that freedom is a postulate although this is in the Analytic. That means that Kant is also aware that he has, in the Analytic, already provided a proof of what will appear as one of the postulates in the Dialectic (and therefore will be presented without any supporting argument) (p. 111).

The cogency of Willaschek’s neat proposal hinges upon his having provided an interpretation of Kant’s notion of practical cognition that shows it to be sufficiently distinct from knowledge, so that it can do the work of making sense of the fact of reason as establishing less than knowledge of freedom, while being sufficient to shore up the postulate of freedom, that is, the necessity to believe in freedom. Willaschek draws upon his recent work with Eric Watkins. There, cognition is differentiated from knowledge insofar as it is a representation of a given object of which the subject is conscious, but without the requirement of assent or any objective justification (Watkins & Willaschek 2017a, b). In this way, it would seem that taking this notion into the practical domain could define a notion of practical cognition suited to the task of addressing Problems I and II. However, the question is: exactly what notion of practical cognition? Indeed, Watkins & Willaschek (2017b) indicated that, while their analysis would focus upon the issue of theoretical cognition, that of practical cognition is more complex.

Willaschek interprets this notion by focussing upon the need for it to leave room for belief which is not knowledge. He draws our attention to the fact that Kant’s ‘typical claim is not that we cognize that we are free, but rather that we cognize the reality of freedom’ (p. 114). Willaschek then draws upon the distinction between reality and actuality. Kant explains objective reality as follows:

If a cognition is to have objective reality, i.e. to be related to an object, the object must be given in some way […]. To give an object is nothing other than to relate its representation to experience (whether this be actual or still possible). (B194–5/A155–6)

So Willaschek is right to point out that reality entails (real) possibility but not necessarily actuality. However, what is the object in question here? In the Fact of Reason passage in the Critique of Practical Reason, Kant asks whether

in an actual case, and as it were, by a fact, one could prove that certain actions presupposed […] an intellectual, sensibly unconditioned causality regardless of whether they are actual or only commanded. (CPrR, AA 5:104)

The object in question is therefore a manifestation of the causality of freedom in action, either an actual action or one that is mandatory. So the reality of freedom does indeed only entail its possibility, but this possibility of acting freely is just what it is to be free. Willaschek wants to distinguish between the reality of a concept and the actuality of its object (the latter is at pp. 114–15), but he mistakenly applies this to the distinction between the reality of freedom and our actually being free while it should be applied to the distinction between the reality and the actuality of free actions (i.e. actions from duty). It is the capacity for acting freely from the moral law that is thereby established, and Kant makes it very clear when he says that ‘that unconditional causality and its faculty, freedom, […] are determinately and assertorically known’ (ibid.): this is not a matter of mere possibility. And indeed, how could it be? If I did not know practically that I am free, the categorical ‘oughts’ of the moral law would become hypothetical with a condition that is unknowable, i.e. whether I am free or not.

My proposal, of course, leaves it unclear why Kant claims that freedom is one of the postulates of practical reason. What is at stake must involve clarifying how practical cognition relates to theoretical knowledge. There is clearly room for ensuring that our conception of practical cognition translates into a requirement of theoretical belief without knowledge, but there is no space to discuss this further here. Much as I disagree with exactly how he draws upon the concept of practical cognition, Willaschek’s focus upon this notion seems to me to be the key to solving Problems I and II.

While Willaschek points out that Kant does not use the expression ‘practical knowledge’, Stephen Engstrom finds that developing such a notion is a useful tool in making sense of Kant’s notion of autonomy. This emerges in a paper where Engstrom tackles first the worry that, if the will imposes the law upon itself, it seems that it could choose any arbitrary law; and second, the concern that it is not clear whether one can make sense of the very notion of self-legislation: if I have an inclination to deviate from the law, I could just annul the latter to avoid breaking any normative constraint. The standard response that the law is formal is open to the Hegelian type of critique that this makes it empty (p. 44–5). Although there are ways of dispatching such Hegelian objections, Engstrom is interested in exploring how to understand autonomy, so as to avoid any such worries in the first place.

Constructivism—i.e. the view that what practical reason does is to construct solutions to practical problems—is viewed by Engstrom as making an important contribution in that it gets us to think of  reason as practical and thus provides an account that explains how the moral law is sufficient to derive norms enabling people to freely live together, it cannot show its necessity (p. 47–8), thus in effect leaving it exposed to a version of the first worry.

While Watkins’s response to the shortcomings of the cognitivist notions of the good will’s conferring goodness upon itself, involved examining how the metaphysical dimension of the notion of condition introduces certain constraints upon what could count as condition, Engstrom views the lack of necessity inherent to the view that moral norms are the result of a constructive process as a problem that is best addressed by appealing to cognitivism (p. 49–50). He thereby stresses Kant’s belonging to a Platonic-Aristotelian tradition as is evident, for instance, from the similarity between the first section of the Groundwork and Plato’s argument that knowledge is the only thing good in itself (p. 50). The main difference with Kant is of course that Kant ends up not identifying knowledge, but the good will as the unconditional good. Engstrom however understands this, not as meaning that Kant’s ‘conception of the good is voluntarist rather than cognitivist, but that his conception of the will is cognitivist rather than voluntarist’ (p. 51). Engstrom is thus emphasising Kant’s identification of the will with practical reason (i.e. the capacity for practical knowledge). Here, Engstrom brackets the problem raised by human imperfection insofar as it is not relevant to understanding autonomy (p. 46). So understanding autonomy must involve understanding what it means to ascribe ‘the self-legislative character of autonomy to the capacity for practical knowledge’ (p. 53).

What follows is a concise investigation into Kant’s understanding of knowledge and the distinction form/content, with the aim of identifying the form of practical knowledge. The text is dense, and one feels that some of it would have benefitted from further exposition and exemplification. Moreover, it is not clear that all the points made herein are needed to address the worries raised by the notion of autonomy. There is no space to discuss these content-rich pages in detail here, but the upshot is that practical knowledge is a knowledge ‘whose subject and object are at bottom identical’ (p. 59). From this, Engstrom concludes that there is a type of legislation that is specific to practical knowledge (pp. 59–60). Indeed, in his analysis of Kant’s conception of knowledge, Engstrom identified a double universality: in respect to the object and to the subject. So, to take his example, in saying that the stone is growing warm, I am claiming both that any stone of identical constitution and in the same conditions would be warming up, and that any subject of identical constitution and in the same conditions would cognise this stone warming up (p. 54). When applying this to practical knowledge, Engstrom can therefore claim that a ‘practical judgment counts as practical knowledge just in case it can be legislated by every person for every person’ (p. 60). The worries about autonomy are therefore allayed through this understanding of Kant’s doctrine of autonomy as expressing the form of practical knowledge: this form is both necessary (so there can be no arbitrariness here) and not empty (so appealing to the mere form of self-legislation is not an empty exercise). In this way, Engstrom has provided a powerful demonstration of the benefit of putting the accent on a cognitivist rather than a constructivist approach to Kant’s ethics.

But is such a moral-cognitivist approach sufficient to make sense of the Formula of Autonomy in the Groundwork, or is there a role for political analogy here? Pauline Kleingeld develops an original take on the Formula of Autonomy (FA) motivated by her observation that, by the time Kant writes the Metaphysics of Morals, he shows signs of having abandoned this third formulation of the Categorical Imperative (CI): autonomy is only mentioned twice there and it is not the FA that is at stake. She claims that the usual interpretations of the FA wrongly view it as understanding autonomy to refer to the agent giving the law primarily to herself, and thereby miss the importance of the political analogy in this formula. She then sets out to show that an important shift in Kant’s political philosophy between the first Critical writings and the Metaphysics of Morals provides the clue to understanding the ‘rise and fall’ (p. 62) of the FA.

Indeed, Kant changed from a view of the just state as one based upon laws that could be accepted by any citizen (i.e. that would pass a universalisability test) to one in which the laws additionally have to meet with actual consent from the citizens (pp. 70–4). According to Kleingeld, the upshot of this change in Kant’s political views is that a formula such as the FA in which the law is made by the agent for all (not primarily for himself) without their consent, would therefore no longer be appropriate if we are to understand the FA in terms of a political analogy.

But that is a big ‘if’: why should we think that the FA must be understood in terms of a political analogy? As Kleingeld herself indicates, in Religion within the Bounds of Bare Reason, Kant explicitly points to the disanalogy between the political and moral realms (p. 75). But, on her reading, this is indicative of a shift away from the position he held when he wrote the Groundwork and the Critique of Practical Reason.

Kleingeld argues that the fact that, with the FA, ‘we are to “consider” or “regard” the will as simultaneously legislating through its maxims’ (p. 64), and that the legislation that is at stake in the FA is primarily directed at ‘all rational beings’ (p. 65). Kant does indeed use these expressions in the Groundwork (GMS, AA 4:431–8), but these are, I would argue, merely his way of describing a counterfactual situation (p. 65). Kleingeld notes this, but still thinks that this can only be understood in terms of a political analogy.

While her discussion of the FA throws up some interesting insights (such as the mistranslation of ‘selbst gesetzgebend’ that does not indicate that the will is legislating to itself, but that it is ‘itself legislating’) there are two main problems with this view. The first is that it would mean that we cannot understand any counterfactual unless we are given a context in which that counterfactual would actually obtain. This is not a plausible understanding of how we understand sentences with counterfactual clauses. If I say ‘If I had been present at a lecture where Kant discussed the FA, I would have asked him about its political meaning’, I do not need to imagine a context in which time travel, my registration at Königsberg university, etc. are a possibility: I just consider an alternative situation.

Second, as Kleingeld herself indicates, Kant ‘states explicitly that several versions of the Categorical Imperative […] involve the use of a “certain analogy”’ (p. 64). So if the purported ‘demise’ of the FA is to be explained by Kant’s dropping the political analogy, then there must be other analogies that are at work in the other formulas. One might point to the analogy of the law of nature used for the Formula of Universal Law (FUL), but to claim that one does not understand the FUL without the Formula of the Law of Nature is not plausible. In fact, universal legislation itself is, for Kleingeld, what requires the political analogy, which means that in saying ‘if one’s maxims were at the same time to become laws for the entire moral community—that is […] that can simultaneously serve as universal laws’ (p. 71) the political analogy is already at work, on her reading. But that means that it is also the case that the FUL requires the political analogy, which defeats Kleingeld’s claim that it is specifically the FA’s need for the political analogy that explains its fate.

The deeper problem, I think, lies in Kleingeld’s apparently believing, or at least giving the impression of believing, that the very notion of moral legislation involves an analogy. When she repeatedly points out that in considering the agent as universally legislating, there is no implication that this legislation is primarily directed at the agent himself (e.g. pp. 62, 65, 66), I would agree, but add that the legislation at stake is defined as the norms that govern an agent’s praxis. That is what morality is about on Kant’s understanding of it. The laws defining my duties in line with the overarching moral law are the laws of freedom (e.g. MS, AA 6:214), which governs my actions when I act from duty; there is no analogy here.

Ultimately, I think one must ask if there really is such a ‘rise and fall’ of the FA. Kleingeld puzzles over its absence from the Metaphysics of Morals. The notion of autonomy itself makes two appearances there, of which she argues that they do not refer to the FA. That is indeed correct, but Kant is thereby indicating that ‘reason’s property of being the source of moral legislation’ (p. 78) is very much alive then as before. So the question should rather be whether the FA is useful to the  purpose of the Metaphysics of Morals of deriving actual duties. Since Kant indicates that the FUL determines the form of maxims and the Formula of Humanity (FH) its material (GMS, AA 4:436–7), the completeness provided by the FA and the Formula of the Kingdom of Ends would seem to make an explicit appeal to the FA as superfluous in the Metaphysics of Morals. And indeed, reading the Metaphysics of Morals, we see that both FUL (e.g. MS, AA 6:453) and FH (e.g. MS, AA 6:462) are sufficient to achieve its aims.

Lucy Allais tackles the very thorny issue of making sense of human beings’ propensity to evil that Kant introduces in the Religion text. Like Kleingeld, she proposes that political considerations must be appealed to here. Unlike Kleingeld, she does not consider their role as lying in defining analogies but rather as characterising the actual situation(s) we are in, and always were in qua human beings (at least when not in a state of nature). Allais lists several features of Kant’s discussion of evil in the Religion (p. 84) and in particular homes in on what I think are indeed two of the most difficult aspects of Kant’s account of evil. The first (Problem A) is, according to Allais, that Kant holds both that ‘an occasional deviation from the moral law’ (p. 84) indicates that we are evil, while a human being is good when she is ‘in a process of “incessant labouring and becoming” (Rel 6:48)’ (p. 84). Since the latter process would seem to imply that there are deviations from the moral law, and since a human being cannot be partly good and partly evil, there is a need for clarification of these seemingly contradictory statements. The second (Problem B) is that although we all have an innate propensity for evil (RGV, AA 6:2, 27–30), this is nevertheless imputable, i.e. this evil is an act of freedom, which again amounts to a combination of claims that are seemingly quasi-contradictory.

Allais starts her investigation by considering accounts of the origin of evil by Allen Wood, Jeanine Greenberg, Seiriol Morgan and David Sussman. Allais agrees with Wood’s proposal to locate the propensity for evil in our social condition (p. 86). She responds to Greenberg’s complaint that doing this undermines the Kantian notion of autonomy, by following Onora O’Neill in interpreting autonomy as being about choosing ‘principles that all could follow’ (p. 87). This interpretation does not involve Kleingeld’s stronger requirement that autonomy be understood through a political analogy, but rather rests on the plausible claim that the social dimension of our human condition be taken into account in considering candidate principles for universal legislation. As long as this social dimension is not understood as defining ‘causal or determining influences on our individual moral choices’ (Greenberg’s worry, ibid.), Allais is right, I think, to endorse it.

There is no space to discuss the other two accounts she reviews, other than to note the importance of Sussman’s idea that our propensity for evil is at least in part attributable to the fact that we develop our rational agency under the guidance of ‘imperfectly rational agents’ (p. 89). While she does not endorse it as such, this clearly plays a role in the position she develops. This involves drawing upon Kant’s observation that ‘when we are not in conditions of justice there are ways in which our actions cannot be right (in Kant’s technical sense) even if we are virtuous and do our duty’ (p. 94). This would mean that while being virtuous is something that is under my control, ‘my relating to others rightfully is not’ (ibid.), something that Allais had discussed elsewhere in relation to the question of begging (Allais 2014). My concern is that referring to a state of injustice that is not under my control cannot explain an imputable propensity for evil: it seems that this is a case where Greenberg’s worry is appropriate.

But although I am not convinced by this way of dealing with Problem B, I think that Allais makes a crucial contribution to tackling Problem A. That is, she correctly notes, following Sussman, that there is an issue with the unity of self that is implicitly assumed in Kant’s statements about being a good or evil agent. Allais finds it puzzling (p. 84). But what exactly is the problem? Kant’s point is that the fact that, at some time, we subordinate the incentive of duty to an inclination and incorporate the latter in our maxim is indicative of what fundamental maxim governs our agency, namely one which allows for such deviations. It is therefore irrelevant whether, at some other time, we incorporate the incentive of duty: we have an evil fundamental disposition (Gesinnung). It is therefore fairly straightforward on Kant’s picture of agency to see that a single deviation from morality signals an evil agent. If it is counterintuitive, some aspect of Kant’s claim might need to be revised. What Allais is suggesting is that there is a problematic implicit assumption here, namely that we are dealing with a unified agent (p. 96). And putting that together with her account of the propensity for evil, she sees our belonging to an unjust social setting as explaining why we might think of ourselves as ‘not properly ordered and unified’ (ibid.). While I am not convinced about this social explanation, I think that she may be right in questioning that we have a unified self. This view would have the further advantage of explaining the idea of the agent’s ‘excessive laboring and becoming’ (RGV,  AA 6:48) in terms of a unification of this self.

Does this involve a revision of Kant’s apparent position about the self, one that would have radical implications for the rest of his theory of agency? It might seem problematic for Kant’s theory of freedom insofar as it posits an intelligible character appearing in the form of an empirical character as accounting for our being transcendentally free. The intelligible character is that ‘through which [the agent] is indeed the cause of […] actions as appearances’ (A539/B567). But Kant does not say that the intelligible character is the only cause of our actions as appearances. Indeed, the human power of choice is ‘pathologically affected (through moving causes of sensibility)’ (A534/B562). This is generally viewed as just the description of action from the point of view of the sensible world. But all such moving causes of sensibility, as appearances (A536–7/B564–5) have grounds that are not appearances. So, our actions are grounded in the intelligible character (which is the crucial point for interpreting freedom) and other intelligible grounds (of the sensible moving causes that determine it, i.e. grounds of natural causality in the various forms it takes in inner sense).

What Kant is describing in the Religion is the question of how the agent prioritises the one with respect to the other. Inclinations are not evil; what is evil is the decision to give them the upper hand. Recalling Engstrom’s discussion of autonomy as self-determination, there are grounds for viewing the evil agent’s self as ‘divided’ insofar as it lets some of these other intelligible grounds of action take precedence over the intelligible character it freely chooses. In so doing, the choice would still be free, but in the sense of a manifestation of the ‘in-capacity’ to act from freedom, i.e. a misuse of the capacity for freedom (RGV, AA 6:29). This interpretation relies upon understanding the intelligible character as the morally good intelligible cause of my actions, on the grounds that the intelligible character’s causality is freedom and that morality is the law of freedom. This is controversial, and there is no space to discuss this further here, but at least, it suggests that Allais’s proposal about the dis-unity of the self can rest on fairly solid Kantian ground. And, as we shall see below, this interpretation can be used to address some related interpretative issues.

Paul Guyer’s paper also tackles the problem of Kant’s apparent inability to account for action that is both free and does not conform to the moral law. A concern with this issue is traditionally associated with Reinhold and Sidgwick, who, so the usual story goes, both found Kant’s presentation of his theory of freedom to imply that the agent can freely only act morally (p. 120–1). Guyer debunks this story, however, by showing that both authors took themselves to be defending Kant against accusations that the latter’s account of freedom was not compatible with allowing both moral and immoral actions to be imputed (p. 121–2). As Guyer explains, the story gets more complicated when one realises that Reinhold was defending Kant against Carl Christian Erhard Schmid’s apparent claim that Kant’s views led to what he termed an ‘intelligible natural fatalism’  (the view that all our action, whether moral or not, is naturally determined). But Reinhold failed to grasp that Schmid was actually interpreting August Heinrich Ulrich’s 1788 attack on Kant’s position.

Guyer then notes that there is a further layer of complexity, and this defines the core of the philosophical part of his paper: in 1792 shortly before Reinhold published the second volume of his papers, Kant published ‘The Radical Evil in Human Nature’, an essay that was to become Part One of the Religion. There, he introduced the distinction Wille/Willkür, i.e. between pure practical reason and the faculty of choice (pp. 122–3) which addresses head on the worry Sidgwick and Reinhold discuss about the apparent incomprehensibility of free immoral action. The Willkür is, as Guyer notes, able to make multiple choices, although these are not in time (p. 124). This picture is one that enables Kant to respond to Ulrich’s criticism that Kant’s account of freedom in terms of the intelligible character leaves no room for freedom of choice because the agent is thereby understood as acting on an unalterable causal ground. It is Guyer’s contention that Kant was always committed to the distinction Wille/Willkür  ‘even if not in those words’ (p. 125). This thesis needs however to be defended against Kant’s apparent repudiation of it in 1797 in the Metaphysics of Morals.

In fact Guyer identifies the source of the problem as lying in the Groundwork. As he plausibly argues (pp. 126–7), one cannot conclude to an identification of practical reason and the will in Section II of that work: Kant clearly describes the will of human beings as ‘not in itself completely in conformity with reason’ (GMS, AA 4:412–13). The problem, according to Guyer, lies in Section III of the Groundwork. Guyer argues that Kant identifies freedom and action governed by the moral law (p. 131–2). Indeed, this step seems necessary if the appeal to our non-sensible nature (as evidenced by our theoretical spontaneity) is to be able to do the work of grounding practical normativity. The problem of drawing upon theoretical spontaneity in this way has been flagged by many commentators (e.g. Guyer, Allison) as problematic so that the argument of Section III of the Groundwork is widely viewed as a failure. But here, Guyer is only interested in the fact that Kant not only identifies freedom with action governed by the moral law (GMS, AA 4:453–4) as just noted, but also then apparently contradicts this claim in the next paragraph (GMS, AA 4:454) by not treating ‘the self in the world of understanding as the ground of the self in the world of sense, but rather the idea of the former self as the ideal for the latter’ (p. 132), i.e. reverting to what Guyer thinks is Kant’s real position throughout his Critical writings.

Guyer has certainly identified an interpretative difficulty here, but there are a couple of issues with his reading that need to be flagged. First, while it is the case that one should expect tensions (or even contradictions) to be found between two different published texts, and one indeed might encounter such tensions within a single large work like the Critique of Pure Reason, an interpretation that claims to have identified a contradiction between two consecutive paragraphs is not immediately appealing. Second, when Kant talks of freedom as the ‘ground of all actions of rational beings’ (GMS, AA 4:452), this is prefaced by ‘in idea’. Exactly how that is to be interpreted is not immediately clear. Third, as Allison (1990) points out, Kant is not consistent in his use of the word Wille: sometimes it is understood as that which is differentiated from Willkür, and is defined as pure practical rationality; but it can also denote the whole of the complex Wille-Willkür.

I suggest that, following upon the interpretation of evil I proposed of Allais’s notion of a divided self, Kant’s argument in the problematic paragraph (GMS, AA 4:453–4) can be interpreted as follows. Kant first defines a rational being’s will as the agent’s causality insofar as she belongs to the intelligible world. Importantly, he then points out that, if I were a purely rational being, my actions would all ‘conform perfectly with the principle of autonomy of the pure will’ (ibid.). This is generally taken as a remark which follows from the previous point, but in fact it is the real major of the syllogistically structured argument. The minor is then the statement, characteristic of Transcendental Idealism, that ‘the world of the understanding contains the ground of the world of sense and so too of its laws’ (ibid.). But that statement is to be taken in all its generality: the laws in question are not just those of an agent’s empirical character, but all the laws of nature.

As a being that is not purely rational, I belong to the world of sense, and as such I am grounded in the intelligible domain (from the minor). My freedom consists therefore in being that ground qua ‘intelligence’, which means that, as free agent, I am subject to the law that governs purely rational beings, i.e. the moral law (from the major). That is, what is the law of the behaviour of purely rational beings becomes a law defining what human beings ought to do, i.e. it is in this sense that I am ‘subject to the law of the? world of understanding’ (ibid.), where ‘subject to’ is to be understood as a statement of practical normativity (this is how I understand Kant’s use of ‘in the idea’ at GMS, AA 4:452) . We can add to this that if I fail to act on the moral law, my action is still grounded in the intelligible (from minor), it is still a causal effect of my will (in the sense of the compound Wille/Wilkür), but there has been a failure of self-determination (I am not autonomous—see Engstrom’s paper): I have freely (in the sense of Willkür) let other intelligible grounds shape its causality.

While this alternative interpretation is no doubt open to objections, it at least has the merit of being in tune with the statements Kant makes in the following paragraph (GMS, AA 4:454) which can be seen to follow directly from this reconstructed argument, by making more explicit in what sense an ‘is’ in the counterfactual situation of my being a purely rational being becomes an ‘ought’ for us human beings because of the grounding relation between the intelligible world and the domain of appearances.

The final section of Guyer’s paper (pp. 133–7) tackles the tricky question of making sense of how to dispel the concern that, in the Introduction to the Metaphysics of Morals, Kant seems to give up his distinction Wilkür/Wille. The worry centres upon the definition of freedom, and the claim that the capacity to choose against the moral law should not be included in this definition (something I have assumed in my interpretation above). Guyer presents a strong defence of this claim based upon an investigation into Kant’s theory of definition in the Jäsche Logik. There, Kant conceives of a real definition as ‘”one that suffices for cognition of the object according to its inner determination […]” (Jäsche 9:143)’ (p. 136). In the case of freedom, the ratio cognoscendi of freedom, i.e. the moral law, must therefore be included and Kant takes it that this is enough. There is therefore no reason, as Guyer has convincingly shown, to doubt that, even in the Metaphysics of Morals, Kant holds that Willkür is clearly distinguished from Wille and enables us to choose against the moral law.

3. Naturalistic Interpretations of the Person

Dieter Sturma wants to change the focus of the discussion of the ‘relation of nature and freedom’ (pp.  138–9) and shift it towards the problem of self-consciousness of which Sturma claims that ‘Kant left behind no fully developed theory’ (p. 138). His interest lies in ‘the phenomenon of self-consciousness and the human life form’ (p. 138). As Sturma sees it, there is a need to shed more light on self-consciousness in particular to help make sense of what Kant’s lyrical pronouncements in the Conclusion of the Critique of Practical Reason (the ‘starry heavens above me’ and the ‘moral law within me’) have to do with the systematic role of self-consciousness (p. 138). As I interpret it, the question is how these statements define a certain kind of self-understanding, although I am not clear Sturma would interpret them that way.

Sturma thinks that the ‘rigidity to which orthodox interpretations of Kant’s dualistic reflections give rise’ should be avoided, chiefly because this has an ‘immense’ negative impact on the ‘attractiveness of Kant’s position for contemporary systematic philosophy’ (pp. 138–9). So this paper has an agenda which is not Kant interpretation, but making Kant more palatable to contemporary philosophers. As I indicated at the outset, it is important for Kant’s work to be present in today’s discussion, so the aim of this paper is, in that sense, a laudable one. But the project seems rather one-sided insofar as it is a matter of adapting Kant, not of looking for points of convergence or issues where contemporary philosophy might do well to reconsider some of its widespread assumptions. This shows a lot of confidence in the success of the systematic contemporary philosophical enterprise.

Two comments are in order here. First, I assume that Sturma is referring to analytic philosophy which is far from being the only paradigm of contemporary thought. Second, the spread of opinions on the proper approach to understanding consciousness within a broadly analytical tradition, with theories ranging from property dualism and (proto-)pan-psychism (Chalmers, Strawson, Goff) to eliminativist materialism (Churchland) and reductive materialism (Dennett), makes it difficult to take the contemporary analytical scene as defining any kind of unity to which the Kantian project might usefully be adapted when it comes to examining issues that are connected with consciousness. Rather, it is arguable that it points to the need to revisit certain basic assumptions, such as the naturalism it assumes (cf. Onof 2008).

As it happens, it is the project of naturalisation as it is carried out in the sciences that Sturma is interested in, with a particular focus upon neuroscience. That naturalisation should be a scientific ambition is practically analytic: the scientist seeks to explain more and more phenomena, whereby these explanations amount to finding a place for these phenomena in nature. That it should be a philosophical ambition is an open question. There is a naturalistic fashion in contemporary analytic philosophy and this is reflected in Kant scholarship. Whatever one thinks of the value of the project of naturalisation in philosophy, it is prima facie puzzling to pick on the Critical Kant to provide support for it: the Critical enterprise is precisely a philosophical revolution that shows the limitations of naturalistic explanations. Now there may be an issue as to where the limits of nature are drawn, but that there are these limits, and that the completeness that reason demands lies beyond them should not be in doubt if the exercise is to be deemed Kantian, or so I would argue. That would be a condition that, I would suggest, needs to be applied to a ‘project of naturalisation’ that is inspired by Kant. A second condition upon any project of naturalisation of the Critical Kant should recognise the following constraint upon explanations: if the possibility of some phenomenon X is to be explained by appealing to condition Y, say, then Y cannot be reduced to some feature of (e.g. internal to) Y. This second condition could be grounded upon Watkins’s understanding of conditioning as a dependence relation with explanatory power, but I would argue that this point can be made independently of such an interpretation.

Before examining Sturma’s proposal in more detail, it is worth adding that if Sturma’s interest had resided in other areas of science, e.g. in interpreting quantum physical theory, he might have taken a very different tack. Some philosophers of science are interested in what Kant has to offer, precisely because of the delimitation of the knowable from the non-knowable, i.e. the existence of what Sturma might consider as ‘too orthodox an interpretation of Kant’s dualistic reflections’ (p. 138). This interest ranges from an interest in the broad outlines of a Kantian approach such as is found in contemporary structural realism (French & Ladyman 2003) to a more focussed interest in the possibilities offered by transcendental philosophy as opposed to naturalism (Mittelstaedt 1994, Bitbol 1998, Bitbol et al. 2003, d’Espagnat 2018).

Be that as it may, Sturma’s paper proceeds by discussing the relationship of nature and freedom (pp. 139–42). Sturma’s naturalistic affinities are in evidence in his interpretation of Kant, for instance when he complains that ‘intrinsic qualities of nature [are] inaccessible for fundamental epistemological reasons’ (p. 140). Is it not rather the case that there are no intrinsic qualities of nature insofar as nature is appearance while an intrinsic property describes the way a thing is in itself? In rejecting this, Sturma’s project is incompatible with the first requirement I identified above. Sturma further singles out what he finds problematic in Kant’s solution to the Third Antinomy in the following terms:

It is not clear, however, how the concept of causality through freedom could correspond to the intelligible character while simultaneously preserving a link to causation. (p. 141)

Since the intelligible character is defined as ‘transcendental cause of the [empirical character]’ (A546/B574), that sentence is strictly speaking incomprehensible. But, again, recalling Sturma’s naturalistic assumptions, it probably means that it is difficult to make sense of freedom as a causality that is in any way connected to the causality of nature. And if that is his point, then I would agree with him, but see that as precisely being Kant’s claim: transcendental freedom is a causality that is to be distinguished from that of nature by definition (A530/B558). Its distinctive feature is that it is a first cause, while natural causality obeys the conditions defined by the Second Analogy: any cause is itself to be viewed as having a further cause (of the event defined as the manifestation of its causality). Sturma however understands natural causality as the only possible causality because he views explanations of nature as having to be entirely internal to it. In this vein, he argues that Kant should have investigated ‘how to situate the constitutive powers of the understanding in a world to which the laws of nature can be successfully applied’ (p. 142). This is, according to my second requirement above, an example of where Sturma’s project of naturalisation can no longer claim to be Kantian.

Sturma does however have some positive points to make about Kant’s project, in particular with respect to the B-Deduction and the Paralogisms: ‘It is Kant’s original insight that the fact of self-consciousness provides us with information about the synthetic power of human consciousness’ (p. 143). His reflections upon the contextualisation of self-consciousness might be of interest to the naturalists he sees as the target for his interpretative efforts. Again, the issue of how Kantian these views are is open to debate. When Sturma claims that ‘the self-conscious person takes up a perspective that allows it to experience itself in the subject-independent world’ (p. 145), it would seem that the first condition I proposed above is not fulfilled: this suggests that the subject’s experience is not thereby delimited, but that she is directly experiencing what it is to be in the subject-independent world, i.e. the world of reality in-itself. One could perhaps object that, on a Strawsonian interpretation of Kant (which is of course controversial, see e.g. Schulting, forthcoming), this is indeed Kantian enough.

A further interesting feature of Sturma’s account is the importance he attaches to the practical perspective that views the subject as ‘act[ing] within the realm of causes […] in relation to reasons’ (p. 145). The distinction he makes between these realms situates his claims within the conceptual space created by the work of McDowell and Brandom, to which he refers. However, here again, the Kantian credentials of Sturma’s claims are questionable. When he claims that the limitations of natural sciences’ accounts of reality ‘do not mean that scientific descriptions as such are incomplete’ and further that ‘the realm of nature […] [is] systematically complete’ (pp. 147–8), he is clearly deliberately ignoring the whole of the Transcendental Dialectic, in line with his decision to free himself of ‘the dualistic rigidity that characterizes many parts of the Critique of Pure Reason’ (p. 146).

I think that Sturma’s project is likely to be of interest to the audience that he is targeting, but I am not clear why he insists on viewing it as a Kantian one. One of his concluding claims, namely that ‘Kant’s critical position’ can be characterised as a ‘naturalism or physicalism’ (p. 151), can only be derived by interpreting away fundamental tenets of Kant’s Critical philosophy.

An interest in naturalism is also what motivates Béatrice Longuenesse’s paper on Kant’s concept(s) of person. But her approach is careful to distinguish Kant’s Critical philosophy from what could be taken from it to inform a naturalistic picture. Her focus is Kant’s Third Paralogism of which she offers a careful reconstruction. She shows how the subject of the major of the paralogistic syllogism is a factual one, i.e. an entity that is identical over time, while the predicate of the minor is the mere consciousness of being identical over time (pp. 158–9). The Paralogism therefore involves an equivocation on this hinge term of the syllogism: it is only if the subject of the major is the predicate of the minor that the argument is valid.

Longuenesse then suggests that there is however ‘a type of entity of which I do have justification to assert the numerical identity at different times’ (p. 162) and which could therefore serve as suitable hinge term for the syllogism. According to Longuenesse, that is ‘a body, endowed with mental states and with the unity of apperception’ of which she says that it would enable me to know my own numerical identity (ibid.). That formulation is ambiguous. That I can know myself as (i) embodied, (ii) with representations in my inner sense, and (iii) as a thinking and apperceiving I, enables me indeed to pull these three characterisations together and take them as determinations of the type of entity that I am. This does not allow me to take any one of these three determinations as defining an entity of which the other two can be predicated. And we can immediately see why: to take the body as what is endowed with the other two properties means that the temporal identity of the body over time would imply a temporal identity of its property of being endowed with an apperceiving I. But just as Kant is considering as logically possible a scenario according to which an ‘I’ moves from one body to another, it is equally logically possible that a given body should be endowed with different ‘I’s at different times. It will be objected that this scenario is not possible for me given that I can recall past states of my body as mine. But how do I know that there is not another ‘I’ which is similarly experiencing states of my body? This is a scenario that is not any more outlandish than Kant’s—and indeed one might think that it is even more plausible to make sense of it, if we take the other ‘I’ to be operating at different times, e.g. when I am not apperceiving.

The problem with this concept of person is analogous to the problem with the rationalist’s conception of person: it is a synthesis of at least two concepts, that of embodied entity and that of apperceiving entity for which no necessary connection has been established, so that the paralogism applies to Longuenesse’s conception as well.

This is ironic given that she then sets out to show that Kant himself is ‘under the grip of his own paralogism’ (p. 163). Longuenesse is thereby attempting to diagnose why Kant does not think that ‘the more modest concept of a person as an empirical entity, conscious of its own identify at different times is not “necessary and sufficient for practical use”—perhaps neither necessary nor sufficient’ (ibid.). Longuenesse notes that, for practical purposes, Kant endorses a notion of person that is close to the rationalist one insofar as it is a pure intelligence, albeit belonging to the intelligible domain and unknown to us. She views this view as standing in tension with Kant’s repudiation of rationalistic conceptions of self and constructs a paralogistic argument that she thinks Kant is guilty of using to support this conception of person. She summarises it as follows:

An entity that is conscious of laying down the law of its own actions is, to that extent, transcendentally free.

I, in ascribing to myself the ‘ought’ of the categorical (moral) ‘I ought to’, am conscious of laying down the law of my own actions.

So I, in ascribing to myself the ‘ought’ of the categorical (moral) ‘I ought to’, am transcendentally free. (p. 166)

Longuenesse concedes that ‘Kant never formulates, nor endorses, such an inference’ (ibid.), but thinks that he makes that move in the Critique of Practical Reason with the doctrine of the fact of reason. There are problems with the premises as Longuenesse presents them. The minor is presumably supposed to correspond to the fact of reason, i.e. the factum argument as Willaschek referred to it in his paper. The fact of reason is an awareness of standing under the moral law. Longuenesse here reduces it to a state of consciousness, but in so doing, she dispenses with the factual aspect of the fact of reason. To be sure, this is not a fact that can be grounded theoretically, but it still defines a theoretical proposition which I have practical grounds to endorse, which explains that it defines a Postulate of Practical Reason, as we saw above. Since Longuenesse claims that it is the absence of factual content in the minor that is supposed to be the key factor in the equivocation that makes the paralogism invalid, there are no grounds for viewing Kant’s use of the fact of reason to ground transcendental freedom as paralogistic.

While one might want to take issue with my account of the fact of reason, there would appear to be a paralogism only if one were to equate the distinction between establishing the normativity of the moral law practically and doing so theoretically with the difference between the mere consciousness of standing under the moral law and knowing that it is normative for rational beings. But that would be to suggest that this difference lies in the possibility of the moral law’s normativity being an illusion, which is surely not a possibility Kant would countenance.

Longuenesse rounds up her paper by providing an interesting development of the conception of personhood that she thinks Kant should have adopted (pp. 169–71). This is likely to be of particular interest to moral philosophers of broadly Kantian persuasion.

4. Kant and Beyond

The last two papers in the volume consider, first, how some key Kantian notions are transformed in Hegel’s philosophy, and second, the interpretative difficulties encountered in examining one of Kant’s last publications, difficulties which also point beyond Kant, though in a different direction.

Robert Pippin turns to the notion of reflective judgement in the Critique of Judgement. His interest lies in drawing upon an understanding of Kant’s bringing together the prima facie disparate topics of aesthetic judgement, the genesis of empirical concepts, the systematisation of empirical laws and our grasp of living nature, to shed light upon Hegel’s notion of ‘act’ of reason in the early, so-called Systemprogramm (p. 194).

Pippin first briefly addresses the ‘vexed and difficult issue’ of the status of mental acts in Kant’s transcendental psychology. A cognitive act is distinguished from a ‘mere “activity”’ such as a computer’s processing because it is norm-responsive. But, Pippin argues, it is also not an intentional action: it is end directed, i.e. teleological, but we are not acting purposively when we perceive, judge, etc. (p. 195).

Next, he draws our attention to the fact that Kant’s distinction between cognitive faculties differentiates functions of thinking, not capacities. Reason is thus the faculty of thinking for its own sake, i.e. ‘pure self-determining thinking’ (p. 196). This is true of pure practical rationality, but also reason in its hypothetical use (see A302/B359). It is to the spontaneity of reflective judgement’s self-determination that Pippin turns his attention to show in what way the activities that Kant brings together in the Third Critique are ‘purposive’ and ‘self-determining’ (p. 197).

Pippin considers Kant’s example of the savage who has never encountered a house (V-Lo/Dohna, AA 24:703) and whose faculty of judgement is seeking to form an appropriate concept for this experience. As he correctly points out, the activities of abstraction, comparison and reflection require that a direction be given to their realisation; how does the savage know how to pick out what will be relevant to future comparisons and is not just a particular feature of this house?

But Pippin’s question is actually two-fold: (i) how is it that any features are salient at all? and (ii) how do we know, among these salient features, to pick out those that will serve to define objective determinations? The Hegelian, conceptualist, response that Pippin is interested in, is that ‘the “ascending” search for a universal cannot begin unless the particular is already determinate enough (and that must mean conceptually determinate) to have a determinate direction’ (p. 198). This ‘mediated immediacy’ does not however provide a satisfactory answer to the first question: since some conceptual salience is, for Hegel, required for further conceptual determination, one is not told how this process of determination could ever get going. As Pippin indicates, Hegel’s notion of ‘mediated immediacy’ is the ‘central and most problematic and paradoxical in his idealism’ (p. 198).

On a modestly non-conceptualist alternative reading (Onof 2016), there are salient features of the environment that are not conceptually determinate, but arise, through sub-personal mechanisms, out of the associations that are constantly being forged through our practical dealings with our environment, and are presented in intuition.[1] These features, which are given in intuition, can be taken by the subject as having objective, although as yet indeterminate, content, and thereby as defining a starting point for the process of comparison, abstraction and reflection that leads to concept formation, i.e. to the identification of objective determinations.[2]

I think however that both this and the Hegelian proposal go beyond what we can find evidence for in Kant’s text insofar as Kant does not address this genetic question. What is however excluded is a strong non-conceptualist reading which would have it that the salient features of our environment are objective features found in the intuitions, independently of any conceptualisation. The shortcomings of this view have been pointed out by a number of authors (Land 2016; Schulting 2015 and 2017, ch. 5).

The second part of the worry is that of the direction of enquiry, and Pippin examines Kant’s answer in terms of the principle of purposiveness of nature. This principle is something that ‘judgment freely requires of itself’ (KU, AA 20:225), i.e. a features of the heautonomy of reason. This notion is, according to Pippin, already present in the First Critique when Kant says that he is interested in what the ‘cognitive faculty […] provides out of itself’ (B1, quoted on p. 199). As such it invites Hegel’s question: why are the categorial structure of experience or the moral law not also what reason requires of itself?

Pippin presents an interpretation of Kant’s doctrine of the higher powers (understanding, power of judgement, reason) that does not place them in opposition to the lower powers (e.g. reason versus desire), but rather views them as the higher forms of the same powers (e.g. reason as the higher form of desire) so that ‘desire is already of the sort as to be responsive to reason, rather than a brute affect to be shared with animals’ (p. 200). This leads him, in particular, to view the notion of respect for the moral law as a manifestation of pure reason’s being practical, namely its sensible manifestation which is not separable from its intellectual manifestation as the imperative to do one’s duty, thereby circumventing typical problems of interpretation of this notion of respect (p. 201). This two-sidedness of our faculties explains Kant’s referring to reason’s needs (e.g. for the unconditioned A314/B370) and Pippin links this to the priority of the practical that accounts for the (practical) indubitability of those matters (God, immortality, freedom) that are beyond the reach of theoretical reason (p. 201).

Pippin’s argument moves quickly here, and is not always easy to follow, but what he has been doing is preparing the ground for a smooth transition to a Hegelian perspective by singling out the conative dimension that constitutes reason’s dynamic as a core feature of the way Hegel presents his argument. Ultimately, for Hegel, ‘reason’s authority is a matter of a heautonomy that is also an autonomy’ (p.  203), so that ‘in knowing itself, what pure thought knows, is the intelligibility, the knowability, of anything that is’ (p. 235). So, while reflective judgement’s principle of the systematicity of nature is for Kant a limitation that reason imposes upon itself, it constitutes the ‘intelligible structure of reality’ for Hegel: ‘To consider being in its thinkability is simply to consider it as it is’ (p. 204–7). In this sense, Hegel clearly moves beyond Kant, but on Pippin’s reading of Kant, the move is one that can be understood as emerging from Kant’s own thought.

Thus, the typical Hegelian account of the movement of the concept, which involves prima facie implausible claims about how the development, for instance, of empirical sciences, can be understood as necessary, becomes plausible if one bears in mind that it is practical contradiction that drives this progress. So, for instance, ‘[i]f a collective attempt to accomplish some goal can be said to learn collectively that commitment to that end is impossible without commitment to […] a broader and more comprehensive end, then it must pursue such a new end’ (p. 208). Ultimately, it is all a matter of reason’s purposive activity which, for Hegel has as its goal a ‘kind of self-knowledge’ (p. 210).

If Pippin’s paper tackles some difficult Hegelian texts to show in what way a perspective rooted in Kant can help make sense of them, Karl Ameriks’s concluding paper invites us to consider Kant’s essay from 1794 entitled ‘The End of All Things’, of which Ameriks says ‘[w]ho knows if anyone has ever really understood it’ (p. 213). Kant uses dramatic language to describe what Ameriks characterises as a proto-Freudian claim that ‘we “cannot cease” turning our “terrified gaze back to it [the end] again and again” […] (EAD 8:327)’ (p. 215). Kant is thereby describing what is ‘another kind of remarkable “fact of reason”’ (ibid.) according to which humans expect an end of the world, and a terrible one.

The notion of end has a dual meaning as temporal and purposive, which Kant emphasises when he makes the proto-existentialist claims that without a final purposive end, ‘creation itself appears purposeless’ (EAD 8:331). Furthermore, even if there is a moral purpose, we are aware that we fall short insofar as we are ‘radically corrupt’ as the Religion (RGV, AA 6:57) has showed, so that ‘it appears that the proper end would be a terribly fitting one’ (p. 217). Kant does not think of this end in terms of Dantesque imagery, but rather a permanent truth that one is stuck to oneself as an evil being’ (p. 218).

After these introductory considerations, Ameriks analyses the essay takes us directly to the focal point of his interest in this paper. Kant describes the end of a person as involving ‘”going out of time” and “into eternity”’ (p. 221), and surprisingly appears to contrast this with a notion of immortality as an infinite progression towards perfection as in the Critique of Practical Reason and other writings. This difficulty is compounded with the claim that beyond time, there will be a ‘noumenal duration’ (EAD, AA 8:327) that is not temporal.

Ameriks presents an interpretation of the notion of noumenal duration that, in its negative dimension, is a notion of eternity as timeless which, as Ameriks indicates, draws on a ‘relatively mainline theological tradition’ (p. 225) that Kant often refers to. He sees Kant as going further though, in giving a positive content to this notion: each person’s imperfect Gesinnung is separated from holiness by the ‘inexhaustibility of the set of steps that rising to holiness would involve’ (p. 226). This idea is directly connected to Kant’s statements about immortality of which Ameriks says that ‘it makes no sense to say that, however peculiar they may look, they are mere slips or pious backslidings or insincere bargains with authority’ (p. 229). How does this define a notion of ‘duration’? Intriguingly, Ameriks proposes a sort of reversal of the procedure of the Anticipations of Perception where we represent qualitative gradations of sensations in terms of a succession of degrees of intensity: sequences of ‘value commitments’ can be represented by, and indeed grounded in a non-temporal ordering of values that expresses ‘degrees of strength of a noumenal will that indicate various levels of possible resistance, within one’s basic disposition, to various stages of temptation’ (p. 226).

This general idea of a noumenal ordering appearing as a temporal sequence seems right, and in line with some of Kant’s claims in the Religion (RGV, AA 6:48—discussed above in the review of Allais’s paper). What is not so clear is how that chimes with the idea of a progression to holiness: it would rather seem to explain how, during our lifetime, moral progression is the manifestation of a certain Gesinnung, but this is an imperfect Gesinnung. In line with what was discussed earlier in this review in relation to Allais’s and Guyer’s papers, it might have been more appropriate to interpret this as a progression reflecting the appearance of our intelligible character (understood as the source of our moral agency, i.e. that whose causality is freedom), while the Gesinnung is just a feature of our temporal existence characterising the extent to which the moral law has the upper hand in determining our agency, i.e. in defining our empirical character. This also avoids the problem of interpreting the Gesinnung as some noumenal reality: since the moral law governs noumenal selves, a less-than-perfect Gesinnung in the noumenal domain is not possible.

Aside from this worry, Ameriks’s proposal leads to a notion of noumenal duration that is somewhat ambiguously poised between ‘a non-perishable mode of being that is not itself within time’ (p. 225) and the claim that ‘metaphorically […] one’s will or underlying character noumenally endures’ in the sense of a potential to appear phenomenally as a phenomenal character (p. 227). This tension is, I think, unavoidable as it reflects the lack of a clear account of how I could ‘be’ my noumenal self without any phenomenal manifestation thereof (i.e. after the end of my earthly life). And here, this volume concludes with Ameriks making a very important (and I think absolutely correct) remark about a problem with Kant’s Transcendental Idealism, namely that the claim of the ideality of time clashes with what Ameriks refers to as ‘primitive and purely psychological temporal aspects of our experience’ (p. 229). This is not to question the ideality of the temporal sequence of events of outer sense, but, as I see it, to point to the essential temporality of any notion of being for the self.[3]

Where does this concern about the ideality of time leave us? It may be that Kant’s focus upon an asymptotic realisation of the Highest Good in a future human species reflects this difficulty, whereby, as Ameriks observes, this is connected with Kant’s realisation that Augustine’s ‘City of God’ represented a vision of a future earthly community, after the Second Coming (p. 230). At the end of this paper that forms a fitting conclusion to an excellent volume in which his influence has loomed large and been explicitly acknowledged by several authors, Ameriks concludes pithily: ‘that is a story some can only hope to tell some other day’ (p. 230).

Received: 9 February 2019. Revised: 23 February 2019.


[1] This proposal has the advantage of explaining cultural differences between the way the objective world is ‘carved up’. An eskimo will have experienced the practical implications of different types of snow which thereby feature as salient differences in intuition, so that the objective concepts of snow he defines have a higher degree of differentiation than those of an inhabitant of Western European flatlands for instance.

[2] To this, it will be objected that these features are subjective, so that nothing warrants taking them as having some, even indeterminate, objective content. I agree that this is a move that goes beyond Kant (see below) but that is only committed to their being some objective content to these salient features: that is, it rather defines as a task the determination of what is objective in such features. Such a proposal goes a step further than Kant in establishing another type of priority of the practical, one that is essentially pragmatic. However, to claim the opposite, namely that there is nothing objective about the salient features that are identified by our practical dealings with the environment, would be counterintuitive.

[3] Although this is too long a topic to address here, some role for a notion of time in making transcendental functions such as the synthesis of a manifold possible: thinking therefore has a temporal dimension that is not reducible to time as a form of inner sense, even though it produces representations that are in inner sense, and therefore temporally ordered. When Kant says that ‘I generate time itself in the apprehension of the intuition’ (B182/A143), this act arguably depends upon a more fundamental notion of time.



Allais, L. (2014), ‘What Properly Belongs to Me’, Journal of Moral Philosophy II (4): 754–71.

Bitbol, M. (1998), ‘Some Steps Towards a Transcendental Deduction of Quantum Mechanics’,

Bitbol, M., Kerzberg, P. & J. Petitot (eds) (2003), Constituting Objectivity: Transcendental Perspectives on Modern Physics (Dordrecht: Springer).

d’Espagnat, B. (2018), Conceptual Foundations of Quantum Mechanics (CRC Press).

French, S. & J. Ladyman (2003), ‘Remodelling Structural Realism: Quantum Physics and the Metaphysics of Structure’, Synthese 136: 31–56.

Korsgaard, C. (1996), Creating the Kingdom of Ends (Cambridge: Cambridge University Press).

Land, T. (2016), ‘Moderate Conceptualism and Spatial Representation’, in D. Schulting (ed.), Kantian Nonconceptualism (London/New York: Palgrave Macmillan), pp. 145–70.

Langton, R. (2007), ‘Objective and Unconditional Value, Philosophical Review 116: 117–85.

Mittelstaedt, P. (1994), ‘The Constitution of Objects in Kant’s Philosophy and in Modern Physics’, in P. Parrini (ed.), Kant and Contemporary Epistemology (Dordrecht: Springer), pp. 115–29.

Onof, C. (2008), ‘Property Dualism, Epistemic Normativity and the Limits of Naturalism’, Philosophy and Phenomenological Research 76(1): 60–85.

——— (2016), ‘Is There Room For Non-Conceptual Content in Kant’s Critical Philosophy’, in D. Schulting (ed.), Kantian Nonconceptualism (London/New York: Palgrave Macmillan), pp. 199–226.

Schulting, D. (2015), ‘Probleme des „kantianischen“ Nonkonzeptualismus im Hinblick auf die B-Deduktion’, Kant-Studien 106(4): 561–80.

——— (2017), Kant’s Radical Subjectivism. Perspectives on the Transcendental Deduction (London/New York: Palgrave Macmillan).

——— forthcoming, Reflexivity and Representation. Essays on Kant and German Idealism (Berlin/Boston: De Gruyter). 

Watkins, E. & M. Willaschek (2017a), ‘Kant’s Account of Cognition’, Journal of the History of Philosophy 55(1): 83–112.

——— (2017b), ‘Kant on Cognition and Knowledge, Synthese (first online).

© Christian J. Onof, 2019.

Christian Onof is Reader at Imperial College London, England, and teaches philosophy at Birkbeck College, University of London. He specialises in Kant, moral philosophy, philosophy of mind and existentialism. He has published in, among other venues, Philosophy and Phenomenological Research, Philosophical Review, Kantian Review, Kant Yearbook, and Kant-Studien. His ‘Reality In-Itself and the Ground of Causality’ is forthcoming in Kantian Review.