I would like to begin by thanking my three commentators, Konstantin Pollok, Jochen Bojanowski and Robert Louden for their detailed and perceptive comments on my book. I am grateful for the time they took in so carefully reading my work. I will not be able to respond to all of the points they raise but will try to respond to some of their objections that relate to the core of my interpretation.

In my book I set out to provide a comprehensive interpretation of Kant’s metaethics that shows both that Kant’s ethics is compatible with a metaphysical naturalism and that Kant is a moral idealist. I take the topics to be interrelated, so that the issue of naturalism is key to some particular aspects of Kant’s ethics while the controversy over idealism is more important to other aspects. The arguments I give for each are mutually supporting and in the end these two main topics in my book converge to form my overall interpretation of Kant’s ethics.

Allow me to first summarize my claims about moral realism so that my critics’ objections, which largely deal with the naturalism aspect of my book, can be put into context. I start by defining moral realism in a way that makes it relevant to the realism/antirealism debate about Kant, where the dividing line is surely not whether morality is valid (true) or not but whether morality is dependent or independent of the moral agent. Moral realists would hold that moral right and moral value do not depend on the moral agent; moral idealists (a term I prefer to ‘antirealist’ or ‘constructivist’) hold that they do so depend. Kant’s approach to philosophy suggests that the dependence relation be understood at two different levels, a transcendental level (understood methodologically not metaphysically) concerning the nature of and requirements for morality and an empirical level concerning actual individual agents. On the transcendental level, the very possibility of morality requires agents who are constituted by their use of pure practical reason in free deliberation and who are therefore obligated by the moral law. Other aspects of morality – I specify eight altogether – are derived from the moral agent’s needs in guiding action by the moral law. There is nothing independent of pure practical reason and thus Kant’s ethics is idealist at the transcendental level. At the empirical level, particular agents can embody this formal transcendental characterization of the moral agent but the content of morality is not dependent upon them; to this extent morality is empirically real because independent of empirical moral agents. Empirical moral agents possess a faculty of reason as part of their minds as studied by empirical psychology. Empirical reason operates as an efficient cause of choices as part of the causal nexus of nature. That same faculty of reason is justified as a valid source of morality at the transcendental level. I also note that Kant’s hesitation to conclude that human beings can go beyond viewing themselves as moral agents to actually proving that they are moral agents allows for the possibility that Kant is a moral idealist at the empirical level; one’s understanding of the practical point of view is also a pivot point between an empirical realism and empirical idealism. My explanation of moral realism and idealism suggests a naturalistic metaphysics for morality in which the empirical moral agent in nature possesses practical reason, the justification of which is methodologically non-naturalistic. I then argue that other aspects of morality that purport to require a non-naturalistic metaphysics, namely free choice, the other postulates, and moral value, can instead be understood entirely within the practical point of view of the empirical agent with no ontological implications. My argument for Kant as a transcendental moral idealist and an empirical moral realist (or perhaps empirical idealist) dovetails with my claims regarding naturalism, and so the book intertwines arguments about naturalism and realism.

Before I turn to my direct replies, I would also like to expand a bit on what Bojanowski has said about my methods and aims. Mine is a reconstructive interpretation that is intended to push the interpretative envelope toward naturalism and realism. I base it on what I take to be the core elements and insights of the historical Kant. In some areas I make claims that, as Bojanowski and Louden note, directly contradict some well-known passages in Kant, but I always do so, as Pollok notes, with some other textual evidence showing that Kant did hold the view I attribute to him in some other contexts or at some other time.

Pollok’s comments focus on his claim that my naturalism could not account for the role of judgement and normativity in ethics. His objection raises questions about two main issues: first, the precise ontology that I attribute to nature and second, the role of the practical point of view of the moral agent, which is grounded in pure practical reason itself, in constituting the normative. The second of these issues touches on the kind of moral idealism I attribute to Kant. Bojanowski’s comments also object to that idealism at this point as one part of his multi-count indictment of my interpretation that draws in my claims about the basis of the moral law in reason, the relation of reason and choice in free actions, and the status of moral value. Louden concentrates on arguing that my account of the postulates as not entailing any non-natural metaphysics is squarely at odds with some of Kant’s most stringent requirements for morality.

I am pleased to learn that Pollok generally agrees with me in interpreting Kant’s ethics to avoid, in his words, ‘any spooky story about metaphysical entities in the realm of ends’ and instead to provide ‘a non-ontological reading of Kant’s practical philosophy’ (Pollok, p. 494). We differ on whether an interpretation that excludes any metaphysical entities and ontology from the practical, one that leaves ontological determination to theoretical philosophy, can be understood as metaphysically naturalistic. My claim in brief is that nothing required by Kant’s practical philosophy involves an ontology that goes beyond a robust empirical realism in which nature consists of empirical objects in space and time, including the mental as well as the physical. I essentially equate the empirically real world with nature. Allow me to put that claim in a broader context. In my book I refrain from any judgement about whether Kant’s entire philosophy, including the theoretical, requires the existence of entities that are not empirically real. I am very sympathetic to the kind of deflationary reading of Kant’s transcendental idealism in which only empirical nature in space and time is real and the transcendental is understood methodologically, and in my book I adopt this methodological approach for part of my argument. But my concern is limited to whether Kant’s ethics can be understood not to require any more ontology than that of objects of empirical nature.

Both Pollok and I take metaphysical naturalism to mean that whatever exists is limited to the objects studied by the mathematical sciences. One must, however, modify this definition somewhat when applying it to Kant because, strictly speaking, Kant allows for mathematical sciences only for matter and not for inner mental life yet he clearly saw mental life as part of nature. Kant thought that mental life could be studied by empirical psychology as a kind of descriptive knowledge that fell short of mathematical precision, and he also just as clearly saw mental life as subject to causal determinism.Footnote ¹ So it is entirely legitimate to understand the object studied by empirical psychology to be a part of nature. Pollok and I agree on this point verbally, but apparently disagree in substance. For I understand the object studied by empirical psychology to be not merely particular thoughts or words that are the outward manifestation of mental activity that we are conscious of in inner sense, but also actual faculties and mental capacities that are understood to ground the particulars that appear in inner sense. Pollok, at least in some places, denies that these capacities or faculties are part of empirical reality – he lists ‘will, intention, choice, decision’ in one publication as alone able to justify rather than explain an action (Pollok Reference Pollok2013: 734) and claims in his comments here that transcendentally free reason cannot be in nature (p. 486).

I believe that all these are in nature, for two reasons. First, the ontology of the mental, while not the subject of much detailed work by Kant, could be comprehensive. Consider for a moment the ontology of appearances in the first Critique. Kant does not limit appearances to actual intuitions or to sensations. Rather, the scope of the actual includes possible intuitions and, more importantly, anything that could be required to explain those intuitions according to laws of nature. Kant gives the example of unperceived magnetic matter that must be actual because required by the laws of nature to explain magnetism: ‘Wherever perception and whatever is appended to it in accordance with empirical laws reaches, there too reaches our cognition of the existence of things’ (A226/B273).Footnote ² Now empirical psychology can work the same way. Although for Kant there are no laws of psychology using mathematics that would allow the same level of certitude, explanations in empirical psychology can include the existence of mental faculties that can ground certain capacities. If some being acts rationally, it makes sense to attribute to that being a faculty of reason as well as a power of choice capable of choosing in accordance with that faculty of reason. Notice that I did not say ‘free power of choice’ – the power of choice as described by empirical psychology would be causally determined either by inclination or by reason, and its freedom could not be understood within empirical psychology but only from a practical point of view.

I argue also that this empirical faculty of reason is in some sense the same as the transcendentally free faculty of pure reason. Putting aside for a moment the doubts that Kant has about whether human beings can do more than believe that they are genuinely rational beings subject to the dictates of pure practical reason (a position I attribute to Kant on the basis of Groundwork III as well as the general nature of the practical point of view), and so assuming that human beings are genuinely full moral agents, my position provides a naturalistic account of the transcendental freedom of reason. I interpret the Third Antinomy to be a discussion not of the freedom of choice, nor of the freedom of individual decision-making, nor of a defence of invocation of reasons or incorporations into maxims, but of a defence of pure practical reason itself as the cause of decisions of the power of choice. In the book I note that Kant’s resolution of the Third Antinomy focuses on reason as a timeless cause whose intelligible character underlies particular decisions: reason as the relevant cause is identified numerous times (A547/B575, A548/B576, A549/B577, among others), and specified as follows: ‘every action, irrespective of the temporal relation in which it stands to other appearances, is the immediate effect of the intelligible character of pure reason; reason therefore acts freely’ (A553/B581). I characterized the resolution to the Third Antinomy as defending the transcendental freedom of reason by showing that the faculty of pure reason, qua faculty with a particular structure and ability, is not dependent upon any sensible causation. The structure of reason is timeless, only its particular manifestations are in time as embodied in particular rational beings.

Pollok questions whether such a conception is feasible. I think there is no more problem with this conception than with a similar view of the faculty of understanding and its pure concepts. Take causation for example. Kant’s arguments are intended to show that any being that will be able to represent to itself an objective experience must be capable of processing intuitions using causal relations to represent the objective order. You and I must embody pure understanding and its concept of cause and effect in our actual cognitive system, and the actual objects themselves as empirically real must also embody causal relations. That our cognitive system arose through evolution does not make the structure of that system and the pure concepts of the understanding dependent upon prior sensible causes, because that abstract structure is justified using Kant’s transcendental argument about the structure of possible experience and of beings who can have that experience. It is of course dependent on natural causes that you or I or any human beings exist, and hence that this particular cognitive structure is embodied, but the structure of our cognitive system is not dependent upon that causal origin.Footnote ³ The same point holds of pure reason with regard to moral experience, which is our experience of deliberation before action, the force of obligation, value assignment and recognition, and the like. Kant claims to have identified and justified the moral law as (the core part of) the structure of pure practical reason that makes possible moral experience (he of course hesitates to assign it the certainty of a transcendental deduction). In this way the faculty of pure practical reason can exist in the empirical mind while nonetheless being independent, qua structure, of natural causation. This is how reason is transcendentally free.

This illustrates a difference that Pollok and I have over the precise scope of metaphysical versus methodological naturalism. At this point Pollok would deny that pure practical reason can exist in nature in this way because pure practical reason is fundamentally a faculty of judgement, and normativity and justification associated with judgements cannot exist in nature. I see this objection as being based on methodological rather than metaphysical concerns. We both agree that Kant is not a methodological naturalist, which I take to mean that Kant understands moral claims to be identified and justified through a philosophical a priori method rather than through any of the sciences, including empirical psychology. One thing that exists in nature, here a faculty of reason in a mind, can be viewed from two different perspectives, first as a result of evolution and of individual growth as a product of natural causation and second as a structure that is justified through a transcendental method independent of any reference to natural causes. This is a methodological rather than a metaphysical difference. The same is true of the cognitive faculty that embodies our ability to process sensations using causal reasoning: any particular person’s embodiment of that ability is the result of individual natural causal antecedents but it is also justified using Kant’s transcendental philosophical method rather than any process invoking the natural sciences.

Still, Pollok wants to deny that this kind of justification can be in nature. He admits that the words and sentences that are used in making judgements are part of the chain of natural causality, but excludes the judgements themselves because these judgements are bearers of validity or praise or blame. I like the way that he pinpoints the issue with reference to my example of a computer performing a mathematical calculation. A computer’s calculation is ‘entirely subject to natural laws’ and so ‘normatively idle’ because only through reason can calculations be seen as correct or incorrect (p. 488). I agree that the computer’s calculation is done according to the laws of nature. But the calculation as a process in nature can also be viewed as embodying the correct mathematical procedure. The determination of the correctness or incorrectness of the answer depends upon the use of the correct kind of process, and that process is determined using a different method than that of natural causation, namely a philosophical (or mathematical or other) method. The computer does not get the correct answer unless the particular subset of natural causal processes involved is the proper one as identified using a method other than mere reliance on natural causality. The process itself is embodied in the physical mechanism of the computer and so is compatible with a metaphysical naturalism. The process itself once identified can be realized in and through an object in nature that follows natural causal laws. One such object is a computer, another is the human brain (or for Kant the human mind as consisting of causally determined mental activity and capacities). The same thing said of the computer holds of the brain or mind in empirical psychology: the process can be embodied in nature but cannot bring validity without being the right kind of process.

So what is still missing? The source of the validity, Pollok insists, is judgement. Talking about uttering words and sentences, he says ‘Only our understanding’s grasp of these words as a judgement allows for assessment and hence validity’ (p. 489). Now I am claiming nothing more than that the individual’s conscious grasping, the act of judgement, does occur in the same object and with no additional metaphysical requirements. The judgement is experienced from a first-person perspective. Empirical psychology could view this same event as part of the causal chain in nature, since an act of judgement is an act of the mind, while of course not endorsing the validity of the judgement qua truth (or moral) claim. The normative element requires a different comprehension of the same set of events in the empirical mind.

Specifically regarding Kant’s ethics, my argument is that reason itself is the source of validity and obligation, which can be understood only from the practical point of view of an agent. The fact of reason is our consciousness of the categorical imperative, and from this agent perspective we are able to impose on nature a separate, moral order of the way that things ought to be. And here is where the naturalism ties to my claims about Kant’s moral idealism rather than moral realism. When making moral claims about moral validity or value, human beings and other rational agents are not referring to some independent order of things but only to reason’s own way of ordering things. As Kant says in the Third Antinomy:

The source of the normative order is reason, but the things that are normatively ordered exist and occur in nature in space and time. In short, the intelligible order of things is not an order of intelligible things but a different order of things in nature. And it is the same reason that is embodied in each particular empirical agent (again abstracting from Kant’s claim that we can prove no more than that we take ourselves to be rational in this way).

Here I turn to a related criticism made by both Louden and Bojanowski. They both contend that moral obligation could not hold if pure practical reason is the transcendentally free faculty but only if the faulty of choice (Willkür) were transcendentally free, which my compatibilist interpretation of the freedom of the faculty of choice does not allow. Louden even tries to turn Kant against me, taking my position to be so similar to a Humean or Leibnizian compatibilism that Kant himself would reject it as a ‘wretched subterfuge’ (Kant 5: 96, Louden p. 478). Bojanowski takes my view of the empirically caused faculty of choice to entail that invoking reason would be merely superfluous: ‘If our actions were fully determined by given desires, the representation of the law-likeness of our action would still be possible, but it would be a mere epiphenomenon that played no causal role in the genesis of our actions’ (p. 468). As should be clear from the discussion of empirical psychology above, I do not claim that our actions are fully determined by desires; rather, in the empirical mind in nature, reason acts as an efficient cause to determine our power of choice. Empirical choice is deemed free when its decision is caused by empirical reason as a motive rather than by inclinations. Both of these empirical faculties in turn are said to have transcendentally free counterparts. I have already described pure practical reason’s ‘structural’ freedom. The power of choice as free in a non-compatibilist, non-natural way is a distinct and for my interpretation a thorny issue. While most of the time Kant says that we are free when our decision is determined by the moral law, Kant does hold in some instances that we must see our choice as not determined at all, or self-determined without any necessitation. I treat this issue in reference to freedom of the will as a postulate in the same way that I treat the other postulates of God and immortality by invoking the status of a postulate as a heuristic device used to support our moral practices. From the practical point of view when faced with a decision, moral agents do act as if they are free in that decision. Whether they actually are or not is irrelevant to their decision-making process from the first-person point of view. Assigning praise and blame is similarly done with the assumption that the agent is responsible though a free choice despite any prior causal influences (although I dispute the importance of this as a transcendental question since Kant often assesses imputation empirically). Upon reflection moral agents use the idea of a non-naturally free choice to satisfy their reason without asserting the truth of the matter. I take this to be the import of the practical point of view. In Groundwork III, Kant certainly accepts that a being who can act only under the idea of freedom is really free from a practical perspective (4: 448). I avoid Louden’s worry that my position is a mere compatibilism because in their moral lives agents are unable to view themselves in any compatibilist way, whereas both Humean and Leibnizian compatibilism would allow for moral agents to view themselves as subject to causal determinism as part of their self-understanding as active moral beings, their practical point of view.

I view the other postulates in a similar way, and Louden’s main objection centres on this interpretation. The postulate of God (I argue) is to be seen as a concept held in the mind heuristically as a way for reason to picture to itself the completion of the highest good. I devote a detailed chapter to showing that Kant’s understanding of the postulates throughout his philosophical career wavered between a view that moral belief was more like a theoretical claim based on practical premises and a view that moral belief was a practical stance or attitude that aided moral action but had no ontological implications. One key point in that chapter that Louden and Bojanowski fail to discussFootnote ⁴ is my emphasizing that Kant has precisely the position that my critics would deny – the use of a concept for action while refraining from taking the concept to refer to any actual object – in his theoretical claim about using the idea of God to ground a teleological investigation of nature. In the Critique of Pure Reason Kant explicitly allows for this use of the idea of God as ‘an object in the idea’ rather than ‘an object absolutely’ that ‘serves only to preserve the greatest systematic unity in the empirical use of our reason, in that one derives the object of experience, as it were, from the imagined object of this idea as its ground or cause’ (A670/B698 and following). My claim is that the postulates can be read with this same status in mind; I show that certainly in the Opus Postumum Kant provides just this explanation of the idea of God in morality, but I offer evidence that it is plausible to attribute the view to Kant even in the second Critique. The postulates can perform their function of supporting morality without an assertion that an object (God), a property of a soul (immortality), or a real non-natural capacity (libertarian free choice) exists that matches that concept or claim. As Bojanowski and Louden both note, I take my reading of the postulates to be an updated version of Hans Vaihinger’s ‘as if’ fictionalist interpretation. I admit that my interpretation ends up denying that we actually have a non-natural transcendentally free power to choose, that we are actually immortal, and that God actually exists, but I take it to be sufficient for a Kantian framework that from our practical agent perspective we are constrained to make our choices as if those claims were true.

Bojanowski raises a different objection concerning my view of the postulates as in some way products of speculative reason (p. 470). In my book I should have been clearer about my claims regarding the nature of theoretical investigation of the practical point of view, and he is right to point out some problems. I do allow that Kant can be understood as investigating the postulates from the point of view of reflection upon moral experience rather than from within moral experience itself, and to that extent I take the discussion of the postulates not to be concerned with the determination of action (the practical) as much as with the explanation of what is (the theoretical). But I claim that this is only one of two ways that Kant understands the role of the postulates. This ‘priority of belief’ interpretation takes the postulates to be items subject to the work of theoretical reason in trying to integrate these beliefs with other theoretical beliefs about the nature of things. The ‘priority of action’, which I see as implicit in Kant and more closely tied to the nature of practical reason, sees the postulates as concepts used not transcendentally but only immanently, not matters of cognition but only concepts ‘unified’ with and ‘only in reference to’ the moral law and the concept of the highest good for the practical purpose of determining action (see e.g. 5: 134). I think Bojanowski’s objection is misplaced when my endorsement of this priority of action interpretation is taken into account.

Moral value is a matter very different from the postulates. I happily endorse Bojanowski’s insightful criticism of the moral value realist that notes that such a moral realist position tries to cross from a theoretical claim about ontology to a practical claim about volition. He and I agree that value can only be practical because it is tied to moral agency. We disagree about whether the link to moral agency still allows for the value of humanity to be intrinsic. I think we agree that the value of humanity is not arbitrary and that the value is merely in the practical perspective. I think we disagree about the status of practical cognition. I take practical cognition to rest on the structure of reason alone, and so I take moral value to be determined by reason at the transcendental level. This sets the standard for each actual moral agent. At the empirical level, I argue that there is no empirically real value property as part of the ontology of nature, and I know Bojanowski agrees. We then disagree about the meaning of the status of the value of humanity as an objective standard rather than an ontological claim. Bojanowski takes this value to be intrinsic to the humanity of rational beings, while I take it to be an inescapable necessary ordering of ends by reason. He argues that our capacity to pursue ends toward something good intrinsically makes humanity valuable; I hold that the value of humanity lies in the role that reason plays in rationally choosing which ends to pursue. In my case the explanation of the value of humanity is that reason is valuing its own realization and practice. Reason is not cognizing any independent system of value. In Bojanowski’s case, humanity is valuable because of some relation it has to the good that is cognized through practical reason. For him practical cognition does not consist of reason constituting the practical point of view but involves, at least in part, reason cognizing some independent value order. My objection to his approach is two-fold. First, what is the independent order if not based on reason’s own functions or some independent ontology? And second, why would any distinct value order be valuable except as some brute fact about things (which I take to violate autonomy)? Bojanowski and others hurl the same objection at the kind of idealist interpretation I and others offer about value: would an imposition of value by reason not simply be arbitrary? My reply is two-fold: first, as mentioned, I take reason to be valuing its own realization and activity, so not arbitrary. My second reply concerns the ultimate justification of the practical point of view itself, but before detailing that I will turn to a related aspect of my interpretation.

Bojanowski objects to my claim that reason ‘produces’ the moral law (my p. 134, Bojanowski p. 469), by pointing out that it is a constitutive feature of reason. Here I think our differences are in part verbal. In using the word ‘produces’ I am stressing that reason does not intuit an independently existing law or standard for the right. Instead the moral law stems from the structure of reason itself and can be said to be a constitutive feature of reason and not a random or arbitrary product. The deeper point raised here, however, is related to practical knowledge or practical cognition. Bojanowski objects to my formulating the idealist position as one in which moral principles are dependent upon the moral agent (p. 465). He suggests that in resting the moral law on the structure of reason that is embodied in the mind in nature I open my position up to charges that I have reduced morality to ‘a figment of the imagination’ (ibid.), the categorical imperative to something ‘not really categorical’ (pp. 464, 467) and Kant to an error theorist (p. 467). Most telling is this passage from his comments: ‘it is not clear why it is this “a priori structure”, and not some other structure, that guides us in our moral thinking’ (p. 465).

My answer to this aspect of Bojanowski’s assessment is to essentially agree with him, albeit without the extreme claim about error theory and figments. And I take the point I am about to make as a crucial part of my claim, one that caps off my idealist interpretation so well that I end the book with it. In several places Kant provides grounds for a claim that he is an empirical idealist who abstains from confirming any transcendental proof of morality, although empirical agents would still view themselves as transcendental moral agents. He argues that there is simply no further explanation for the faculty of reason itself, for why it has the structure it has rather than some other: near the end of the resolution of the Third Antinomy Kant says that the question ‘why reason has not determined itself otherwise’ is incoherent, and the question ‘why reason has not determined appearances otherwise’ is unanswerable because it depends upon a successful answer to the first question. Asking why reason has the intelligible character (structure) that it has ‘surpasses every faculty of our reason, indeed, it surpasses the authority of our reason even to ask it’ (A557/B585). At the end of Groundwork III Kant links practical philosophy to this inability to find a final explanation when he says of reason’s search for the necessity of the moral law that it is an

I take it that what Kant means here is that we must simply accept that there is an ultimate inexplicability at the foundation of our moral lives. The only way we would be able to fully explain things is through reason, but the nature of reason is that it seeks an ultimate explanation, and such explanations are impossible. This is similar to the way that the theoretical ideas of reason function to satisfy reason’s search for an unconditioned without reason needing to affirm their reality. Further, in practical philosophy it is reason itself that we would need to explain since reason is the source of the moral law. We human beings are forced simply to accept the categorical imperative presented to us as a fact of reason. There is not only no transcendentally real moral order external to the transcendental moral agent, there is also no guarantee that the moral order we receive from reason has any basis other than the faculty of reason we happen to have. Reason provides itself with a self-justificatory picture including not only the claims about the transcendental structure of moral agency but also the postulates of practical reason and the imposition of a value structure on nature. Since there is no way to confirm the ultimate validity of reason’s claims, we are left with what I label an empirical moral idealism in which our existing mental structure in nature provides us with an inescapable framework for our active lives. The very source of that framework, our actual human reason, provides a comprehensive but unverifiable justification of itself that constitutes (in part) the practical point of view. And, to quote the final sentence of Kant’s Groundwork, ‘this is all that can fairly be required of a philosophy that strives in its principles to the very boundary of human reason’ (4: 463, my emphasis).