2. The FH

But what would it be, in any case, for practical reasons to enjoy explanatory pride of place in a theory of the rightness and wrongness of actions? There are, inter alia , both realist and constructivist answers to these questions, which might be formulated in terms of corresponding treatments of Kant’s FH, to which I now turn.Footnote ³ Kant’s remarks in the transition from the FUL to the FH in the Groundwork of the Metaphysics of Morals have been seized upon by realist and constructivist readers alike. Here he claims that if “there were something the existence of which in itself has an absolute worth,” then “in it, and in it alone, would lie the ground of a possible categorical imperative.” He then claims that there is such a thing: “every rational being exists as an end in itself.” He says that, unlike “objects of the inclinations,” and “[b]eings the existence of which rests…on nature,” which “have only a relative worth,” rational beings are “objective ends,” without which “nothing of absolute worth would be found anywhere,” so that “no supreme practical principle for reason could be found anywhere” [ Grundlegung 4:428].

Allen Wood, who will serve as my paradigm FH-realist, is moved by Kant’s claim to the effect that if there is a categorical imperative, its ground is an end in itself. He takes this to mean that the categorical imperative, as a principle of the rational will, is grounded in a kind of value that some beings—beings with practical reason—just have. He takes Kant to argue that this status is a condition on the value of any of our ends: all conditional worth must have its condition, so that if you go far back enough, you must reach something with absolute worth (Wood, Reference Wood2008: p. 92). So as long as you accept that your ends have conditional value, if Kant’s argument works, you must accept that you and all other rational beings have the status of ends in themselves.

Christine Korsgaard, who will serve as my paradigm FH-constructivist, agrees that rational beings have a status that conditions the value of anything else and that this status is the ground of the categorical imperative. But Korsgaard does not think that anything just is valuable. Rather, we value things and in so doing confer value on them. And insofar as we confer value on anything, we confer, or are committed to conferring, the status of end-in-itself on ourselves: “[y]ou must value your own humanity if you are to value anything at all” (1996: p. 123).Footnote ⁴ Both Wood and Korsgaard thereby endorse regressive readings of the transitional remarks, as expressing an argument that concludes that something—rational beings—has a status which is, in turn, the explanatory ground of the moral law. But they understand this status in different ways: in a value-realist way, in Wood’s case, and in a value-constructivist way, in Korsgaard’s. The difference comes to this, that Korsgaard takes our value to depend upon some act that we perform or are committed to performing, whereas Wood denies that there is any such dependence (Wood, Reference Wood2008: p. 92).

But it is not obvious that Kant holds the value of humanity to be the explanatory ground of the law in either of these senses. Consider the following alternative reading of the transitional remarks. We know that the law says to act only on that maxim which you can at the same time will that it should become a universal law. But what is it so to act? How can we describe this kind or mode of action in such a way as to make it both motivationally and normatively intelligible that anyone would ever so act? Kant’s answer is that to act only on that maxim which you can at the same time will that it should become a universal law is to treat all humanity never merely as a means but always as an end. To say this is to bring the law “closer to intuition” and gives us “access” to it [ Grundlegung 4:437]. Thus, on this proposal, humanity’s status as an end in itself is not an explanatory ground of the law—a fact that explains why we are bound by it—but rather the meaning and spirit of the law and thereby elucidates what we are doing when we act from it.

But suppose we say that, in preference to the positions of Wood and Korsgaard. Then what are we to say about the explanatory ground of the law in practical reason? If it is not the value (or valuing) of practically rational beings, what could it be? It could be practical reason, or the rational will, itself: practical reason itself, rather than any value it might be said to have. My own preference is for a constitutivist specification of this idea on which the law articulates an internal standard of the will’s exercise, so that to the extent that you act as the law prescribes, you exercise your will well, and to the extent that you do not, you exercise it badly.Footnote ⁵ When Kant says that an imperative is “possible” [ Grundlegung 4:417], I understand him to claim that it expresses an internal standard for volitional activity. And that is why it is binding on volitional activity. Evidence for this is to be found in the character of the arguments he furnishes for his possibility claims, both in the case of the hypothetical imperative—if you do not take sufficient means to your ends, you fail on your own terms, because to set an end is to will to make yourself the cause of something, and taking sufficient means is nothing other than making yourself the cause [ Grundlegung 4:417]—and in the case of the categorical imperative—if the maxims you will are not universalizable, you fail on your own terms, because to will a maxim is to institute a law for the rational will, but only if your maxim is universalizable can it be a law [ Grundlegung 4:446–7].

Now, I have just opposed the constitutivist conception of the groundedness of the law, claiming it as my own preference, to the value-realist and value-constructivist views of Wood and Korsgaard. But, funnily enough, this constitutivist story is one which Wood and Korsgaard themselves explicitly affirm, namely, that the law is grounded in the nature of the will, and in precisely the sense I mean. Both of them explicitly endorse a model of the relation of the law to the will in, as Wood puts it, the terms of “the natural teleology involved in…proper functioning” (Wood, Reference Wood2008: p. 115); see, for a similar characterization, Korsgaard (Reference Korsgaard2009: p. 59ff.). But if I have billed constitutivism as an alternative to value realism and value constructivism, how could each of them endorse both it and one of these other positions?

As far as I can tell, neither Wood nor Korsgaard gives a clear answer to this question. The most straightforward answer on Wood’s behalf, it seems to me, is that it belongs to the will’s function to observe a value independent of this function, not in the sense that there would be anything with this value were there nothing with this function—since it is only things with this function that are said to have the value in question—but in the sense that what it is for them to have this value is not exhausted by the fact that it is the function of the will to treat them in a certain way (contrary to the proposal of, e.g., Sensen (Reference Sensen, dos Santos and Schmidt2018)). In Korsgaard’s case, there are two alternatives that her texts seem, at first glance, to make available. On the first, her value constructivism is actually not explanatorily additional to her constitutivism after all. Rather, she holds that part of what it is to exercise the will well is to value humanity in yourself and everyone else. On the second, like Wood, she holds that it belongs to the function of the will to observe a kind of value independent of this function in the sense that what it is to have this value is not exhausted by the fact that it is the function of the will to treat beings with it in a certain way. But she has a constructivist, as opposed to a realist, conception of this value, in the sense that she thinks it is conferred or constituted by acts of valuing or by our commitment to performing such acts. These two Korsgaards thus endorse different orders of explanation. The first Korsgaard says that you are committed to valuing humanity because the internal standard of the will tells you to value it. The second Korsgaard says that the internal standard of the will tells you to value humanity because you are committed to valuing it.Footnote ⁶

I draw out these speculative interpretive possibilities in service of the following point. It is easy to see why Wood, given his value-realist explanation, might insist, against the second Korsgaard’s value-constructivist explanation, that Kant is a realist and not a constructivist—this notwithstanding the constitutivism that they both endorse. But, surprisingly, it is not merely on the basis of the positions on which they disagree that they come to different conclusions on the realism-versus-constructivism question but also on the basis of their common constitutivism. That is, from the claim that both Korsgaard, Wood, and I all accept—that the moral law is grounded in the nature of the will in the sense of expressing its internal standard—and from that claim alone, Korsgaard and Wood draw opposite conclusions concerning Kant’s metaethics. While Wood thinks that Kant’s depiction of the law as grounded in the nature of the will makes him a realist, Korsgaard thinks that it makes him something else.

How could Korsgaard and Wood reach the opposite conclusion on the strength of the very same claim? Because, I believe, they disagree about what it takes to be a realist. Wood emphasizes that if the law is grounded in the nature of the will, then, more generally, it is grounded in the nature of something rather than any act I perform or attitude I have. And that, he believes, is sufficient, in the terms of contemporary metaethics, for Kant’s position to be realism (Wood, Reference Wood2008: p. 113). Korsgaard, on the other hand, appears not to believe this. She appears to believe that metaethical realism requires the constitution of moral facts, properties, or values independently of the will’s nature, not just independently of any of its exercises.

3. The FA: Or, Systems of Heteronomy

In what remains I shall explain why, notwithstanding that Kant thinks the law is grounded in the nature of something—as Wood and Korsgaard both believe—it matters a great deal—contrary to Wood’s judgment, and as Korsgaard would insist—which natures, and how many natures, we have in mind. I shall explain why this matters even if we reject the second Korsgaard’s value-constructivist explanation of the will’s function and, in general, any position according to which an act or attitude of the will plays an explanatory role in connection with the normativity of morality.Footnote ⁷ For in my view, the Rawlsians are right that there is a difference between the rationalists and Kant which realist readers underplay. Kant himself has a name for this difference. As he formulates it, while his rationalist predecessors endorse systems of heteronomy, Kant’s ethics is a system of autonomy.

There are two major discussions of the autonomy/heteronomy contrast in Kant’s published ethical works. One is in the Groundwork; the other is in the second Critique. In both discussions, Kant lays out a taxonomy of systems of heteronomy meant to include various ancient and modern moral psychologies. The principal division is between sentimentalist and rationalist conceptions of moral motivation. Conceptions of the sentimentalist kind he regards as obviously systems of heteronomy because, roughly speaking, according to them, I am morally motivated just insofar as there are some feelings with which I happen to find myself saddled. These feelings are not representations of anything normative for my action: they are not perceptions of the good or of duty. When I am motivated by them, therefore, I am not acting as I do because it is the right thing to do; I am merely chugging along in accordance with the idiosyncrasies of my psychological constitution. Rationalists may seem, on the other hand, to hold a view as different from this as any view could be. They say that I am motivated by my representations of the good, by an independently constituted normative order. When I am acting as I do, it is precisely because it is good to do so. And yet, as I shall show, Kant claims that this apparent difference between the sentimentalist and the rationalist is an illusion.

There are, I believe, several interconnected strands of argument for this claim to be found in the texts I have in mind, some of which I shall address in Section 5. For the moment, I wish merely to begin to state Kant’s position with the aid of this exemplary passage:

In this passage, we are told that in a system of heteronomy, the special constitution of the “nature” of the subject, whether this arrangement is sensible or intellectual, means that its powers “employ themselves with delight upon an object.” This suggests the following model for heteronomy: there’s some object; I am naturally so constituted as to take delight in exercising my powers on it; for this reason, the object grounds laws for me. Characteristic of heteronomy is the explanatory priority of the object to my orientation towards it. More concretely, according to the moral heteronomist, there is something called the good, and I am so constituted as to want it and pursue it; I am, as we might put it, a lover of the good.

If the priority of the object to my orientation to it is characteristic of a system of heteronomy, then characteristic of a system of autonomy must be the opposite. It must be that, according to a system of autonomy, no explanatory significance is accorded to anything defined and constituted independently of the will, and in particular no explanatory significance is accorded to anything called “the good” defined independently of the will. And indeed, that is Kant’s position. His notion of the good is precisely not the notion of something defined and constituted independently of the will, to which the will is then said to be oriented. Rather, the Kantian good just is the object of the will. By “the good” “is understood a necessary object of the faculty of desire” [KpV 5:58]; “the concept of good…must not be determined before the moral law…but only (as was done here) after it and by means of it” [KpV 5:63]. Our relation to the good is not a relation to a normative order lying outside our own nature.

And with that, I can give a preliminary statement of the positive part of my verdict on the Rawlsians. This idea, that Kant denies that our relation to the good is a relation to a normative order lying outside our own nature, is that in which Kant differs from his rationalist predecessors. I submit that the spirit of this idea is present in much work in the Rawlsian interpretive tradition and indeed in something that Rawls himself emphasizes both in his 1980 paper on Kantian constructivism and in his Lectures:

Of course, there is an important difference between the contrast I have drawn and the one Rawls here draws. Mine concerns what depends, and what does not depend, on our nature. His concerns are what depend and what do not depend on a conception: on the conception of the person in particular. But my point now is merely that the topic of this passage is the question whether we are to be taken to be connected to a normative order constituted independently of us. Its topic is not a procedure or the construction of value in valuing.

4. Two Forms of Constitutivism

But how exactly should we label this distinction between Kant and his rationalist predecessors? Rawlsians have traditionally called Kant a “constructivist,” but—here I certainly agree with the realist readers—that really does not seem to fit. So how about one of the other terms that tradition has produced? I myself have indicated a preference for a constitutivist reading of Kant. So “constitutivist” seems, as a label, a good candidate.

But actually, the matter is more complicated than it may first appear. And that is because, though I indeed believe that Kant is a constitutivist, I do not believe that this is what distinguishes him from his rationalist predecessors, since I believe that at least some of them are constitutivists too. The distinction between Kant and his rationalist predecessors is one between two forms of constitutivism—a distinction that the more recent Rawlsians, who make use of the term “constitutivism,” have not acknowledged. Indeed, Korsgaard claims that, as she puts it, the “combination of [a] rationalist view of normativity with a formal principle of reason is unusual” (Korsgaard, Reference Korsgaard2009: p. 64). But if by “formal principle” she means “internal standard” and by “rationalist view of normativity” she means an order of, as she sometimes puts it, “eternal normative verities” (ibid.), this combination is common enough, historically speaking. Korsgaard considers a position on which reason is “a normative module that has been inserted into you, for the purpose of making the laws of reason, which are essentially outside you, also be a force within you” (Korsgaard, Reference Korsgaard2009: p. 6). It is clear that she does not interpret this position as boasting the explanatory advantages of constitutivism. For it leaves open the skeptical question of why we should “conform to the demands of this little representative within us” (ibid.), whereas constitutive standards “meet skeptical challenges to their authority with ease” (Korsgaard, Reference Korsgaard2009: p. 29). This seems to be because constitutive standards apply to something by virtue of what it is, whereas the normative module, being merely inserted into you, is not what makes you what you are. But pre-Kantian heteronomist constitutivists believe precisely that your orientation to a normative order outside of you is (part of) what makes you what you are.

In particular, many of Kant’s rationalist predecessors will agree that the standards of success in the exercise of the will are explained by the nature of the will: that is, that they are internal standards. This is different from another kind of position, according to which the nature of the will does not explain the standards for its exercise, but that there nonetheless are such standards. Good candidates for externalist readings include anyone with a purely elective conception of choice who endorses the doctrine of the liberty of indifference: Karl Leonhard Reinhold, for example, who famously complains that Kant is an externalist about choice (Reinhold, Reference Reinhold1792: pp. 271–277). As I understand constitutivism, it is generic enough a notion that the differences between, say, Aquinas’s more intellectualistic and Duns Scotus’s more voluntaristic accounts of the relation of the will to the good reflect different implementations of constitutivism because in both cases, the good is explained in terms of the perfection of the will and thereby in terms of an internal standard.Footnote ⁹ Scotus, for example, writes that “the natural goodness of a volition…characterizes it insofar as it is a positive being—which indeed characterizes every positive being according to the degree of its being: the more being, the more goodness; the less being, the less goodness” [ Ordinatio II d. 7 q. un., nn. 28–39 (Duns Scotus, Reference Scotus and Wiliams2017: p. 123)].Footnote ¹⁰ William of Ockham, on the other hand, may plausibly be read as endorsing no such implementation and as holding that there is no explanation, in terms of what the human will is, of the normative demand that it be related to the good. Thus, as Calvin Normore puts it, he may be read as divorcing the “action-guiding” sense of “good” from the “metaphysical” sense (Normore, Reference Normore1998: p. 35). In general, if Ockham is to be read, as Normore suggests, as embracing an extirpation of internal teleology from created essences, then he is equally to be read as rejecting a constitutivist account of any normative standards applicable to creatures.

Constitutivists are known to complain that externalists cannot explain how external standards can be normative for the will, and this is a complaint that Kant would share. But it is not his complaint against his German rationalist predecessors, Leibniz and Wolff, because they are not externalists. They embrace internal teleology and follow figures like Aquinas and Duns Scotus in thinking that there is something in the nature of the will that orients it to the good and thereby constitutes the good as its internal standard. Leibniz, for example, writes that “the will is never prompted to action save by the representation of the good” [Theodicy 45] (Leibniz Reference Leibniz and Huggard1985) and, indeed, in commenting on the relation “of the will to the good,” that “the effort to act in accordance with the judgment” of what is good “forms the essence of the will” [Theodicy 311 (Leibniz Reference Leibniz and Huggard1985)].Footnote ¹¹ Wolffian perfectionism might be understood as a development of this Leibnizian idea: the will, for Wolff, is the capacity to promote perfection (e.g., Wolff (Reference Wolff1720: pp. 10–11 et passim )).Footnote ¹²

So Leibniz and Wolff, like Kant, can be read as constitutivists, as holding that exercises of the will are successful to the extent that they satisfy the will’s internal standard. Nonetheless, they disagree with Kant about whether that is the end of the explanatory story. He says “yes,” they say “no.” Leibniz and Wolff say, in their different ways, that the will is the capacity to pursue the good, where “the good” is defined in normative terms independently of the will. Kant, in saying what the will is a capacity to do, does not use any independently normative notions. This is what distinguishes his autonomist constitutivism from their heteronomist constitutivism.Footnote ^13, Footnote ¹⁴

And with that in hand, I can state my final verdict on the Rawlsians. What they have right is not that the rationalist thinks that the moral law is grounded in the nature of things, whereas Kant thinks that it’s grounded in some act or attitude. Kant agrees with the rationalists that it’s grounded in the nature of something. Nor even is it that while Kant thinks the law is grounded in the nature of the will, the rationalist thinks it’s grounded in the nature of something else. Leibniz and Wolff agree with Kant and disagree with figures like Ockham on this point. What they have right is that, for Kant, unlike for the rationalist, there is no normative order independent of the nature of the will in terms of which its internal standard is to be understood in turn. For Kant, the explanatory buck stops with the nature of the will in a way in which it does not for the rationalist and her medieval antecedents.

5. Kant’s Arguments

At this point, I claim to have done what I promised, which was to say what is right and what is wrong in the Rawlsian reading, or at least in its spirit. But one thing I have not done is to consider Kant’s arguments for autonomist constitutivism. This section will include the sparest sketch of some of these arguments, intended not to eventuate in a judgment in his favor but rather merely to sharpen the description of his position.

In the passage from the second Groundwork considered above, there are two reasons given for Kant’s position: first, that only autonomist constitutivism accounts for the a priori character of our knowledge of the good; second, that only autonomist constitutivism accounts for the categorical character of moral requirement. These two reasons are, moreover, easy to construe as a package deal, because he thinks that moral requirement can be categorical only if our knowledge of the good is a priori [e.g., Grundlegung 4:389 f.]. I myself doubt that he can make a convincing case for that claim or that he can show that none of his predecessors can account for categoricity.Footnote ¹⁵ His basic thought is that for them, moral principles can only be hypothetical imperatives, because they instruct the agent to do something to which she is oriented by default. But I suspect that this criticism betrays a lack of appreciation for the structural complexity of the positions to which he opposes his own. To take one example, Aquinas says that in its default orientation, the will is directed to the good of its possessor; it is only when it possesses the virtue of justice that it is equally directed to the good of others [DQV 5c; Aquinas (Reference Aquinas, Hause and Pasnau2014: pp. 522–523)]. So moral commandments are not hypothetical imperatives relative to the agent’s default orientation. Of course, what happens when the will has its virtue will be described by Aquinas and Kant in very different ways, but that is a further issue. That a notion with the structural features of obligation can be constructed for a heteronomist constitutivist, along the lines I have suggested, seems tolerably clear. If it cannot, Kant himself does not do the work of showing this.Footnote ¹⁶

Thus, I take leave to set aside the issue of categoricity and to ask whether independent arguments can be given which focus on apriority. I begin by operating at one remove from the text, to which I shall return presently. So here’s a Kant-style argument to the effect that the heteronomist cannot account for the apriority of our knowledge of the good. The heteronomist says that we human beings are lovers of the good, which is something constituted and defined independently of us. But how does she know that? To know that, you have to know three things: human beings, the good, and a relation between them. But how does she know the good? Ex hypothesi , not empirically. So does she have an intellectual intuition of the good? Or is this rather to be understood as a theory that we should accept on the strength of its various virtues by application of inference to the best explanation?

You might like some of these epistemological alternatives. Then again, like Kant, you might not. You might think that the idea that we are possessed of intellectual intuition is enthusiasm akin to dreams of a spirit seer. You might think that assessing theories according to their virtues has no place in philosophy, in which hypotheses are contraband [Axv]. You might agree with Kant, on the positive side, that to the extent that we have a priori knowledge of something, that something cannot be constituted independently of us.

This essay is not, of course, the venue for a comprehensive examination of Kant’s arguments for this positive claim about a priori knowledge. But it will help in the present context to draw out some of the attractions of this position on the a priori in application to the case of morally motivated action.Footnote ¹⁷ This will also allow me to get to the explanation I have promised of the sense in which the sentimentalist and the rationalist are both perspicuously called heteronomists, notwithstanding that their positions initially appear to be as different from one another as any two positions on the same issue might be, the sentimentalist asserting that I do what I do on account of the idiosyncrasies of my constitution and the rationalist asserting that I do what I do in observance of an external normative order.

Recall that the heteronomist characterizes the will in the independent terms of the good: the will is the capacity to pursue the good. I exercise the will successfully to the extent that I do so in pursuit of the good. But now let us consider another theorist of the will, one who takes issue with our heteronomist’s characterization of its nature. Our new theorist says that the will is a capacity to pursue not the good but the gluttonous consumption of biscuits. She is not an agathist but a bisculinarist. The bisculinarist says that to the extent that I pursue the gluttonous consumption of biscuits in the exercise of the will, I exercise it successfully; and that to the extent that I do not, I exercise it unsuccessfully.

Doubtless this conception of the human will makes for an image that is ridiculous rather than sublime. But notice that, as far as the constitutivist explanatory machinery is concerned, it works in exactly the same way. Success and failure of exercise are determined by the nature of the capacity; it matters not whether the nature of the capacity is spelled out in an infinitival expression that makes use of independently normative vocabulary. That you succeed is just a matter of satisfying the internal standard of the capacity you exercise; that you fail is just a matter of not satisfying it.

This observation is what leads us to the problem with heteronomist constitutivism, from Kant’s point of view, but to see how, it will help to taxonomize one more time, and divide the sphere of heteronomist constitutivism in two: for the argument must be run differently for each of these two varieties. The heteronomist constitutivist says that the will is the capacity to pursue the good, where the notion of the good is an independently normative notion. This entails that the will is the capacity to φ, where φing is equivalent to pursuing the good. But now we may distinguish between a heteronomist who says that the nature of the will itself takes account of this equivalence—so that the will is oriented towards the good as such—and a heteronomist who says that it does not, so that there is a mere harmony of the will and the good, though no direct explanatory relationship between them. There is no difficulty in characterizing a sense in which the latter heteronomist cannot claim that the acts of our will observe an external normative order. According to her, when I exercise my will well, I φ because φing constitutes success in such exercise; that φing is equivalent to pursuing the good is, by her own admission, neither here nor there.

It takes a more extended argument to characterize a sense in which the former heteronomist—who is more to the point, being the type of those of Kant’s predecessors I have chiefly in mind—cannot claim that the acts of our will observe an external normative order. First we need a more detailed characterization of some features of the will. The will is a desiderative or appetitive power, a power of knowledge the object of whose act is existentially dependent on the act itself. It is also a self-conscious capacity, in the sense that it admits of a kind of exercise that contains and is explained by its bearer’s knowledge of that exercise as successful: that is, as meeting the capacity's internal standard. For the bearer to know, internally to the exercise of the will, that this exercise meets the will’s internal standard is a fortiori for the bearer to know, internally to the exercise, the standard itself and thereby the nature of the will as a standard-governed capacity. In such cases, then, the will’s act is knowledge both of what depends existentially on this act and of something—the standard of exercise—which is prior to the act but not prior to the nature of the will.

And this, moreover, exhausts what can be known internally to the exercise of the will. For apart from what its exercises bring into being and its own nature, all that there is to know is that which is prior both to the will’s acts and to its nature. But the will cannot be a capacity to know any such thing, for it is not a cognitive or apprehensive power but a desiderative or appetitive power. And this is what entails that there is a sense in which the heteronomist, even if she embraces a direct explanatory dependence of the will’s nature on a normative order outside it, cannot claim that its exercises can observe this normative order. She cannot claim, namely, that these exercises contain knowledge of themselves as in agreement with this normative order as such, because they cannot contain knowledge of that order as such. Thus, even for this kind of heteronomist, when I exercise my will well, though in one sense I do what I do because an external normative order demands it, no knowledge of this explanatory relation is internal to my act. Insofar as what explains my act is knowledge internal to the exercise of the will, it is knowledge merely of the conformity of the exercise to the will’s standard. That the latter is explained by an external normative order is neither here nor there. The significance of the difference between the sublimity of agathism and the ridiculousness of bisculinarism escapes the self-consciousness of my act of volition as such.

And this is the sense in which the rationalist and the sentimentalist are, from Kant’s point of view, of a piece. The sentimentalist says that I do what I do not because of the demands of an external normative order but merely on account of the contingencies of my constitution. The heteronomist constitutivist wants to say that I do what I do in recognition of an external normative order, but the structure of her position means that she is forced to agree with the sentimentalist. Sentimentalism and constitutivist rationalism both have the consequence that I do what I do, not because it is (externally) good, but—insofar as this is a self-conscious “because” internal to the exercise of the will itself—merely on account of the contingencies of my constitution. This is the real reason why, as Rawls repeatedly insists, Kant rejects moral rationalism “as a form of heteronomy just as firmly” as he rejects “Hume’s psychological naturalism” (Rawls, Reference Rawls and Herman2000: p. 235).

Now, having introduced that argument at two removes from the text, I want to return to the text—but not to the text of any of Kant’s practical works. I want to consider something in the Critique of Pure Reason, something that he says about the categories of the understanding. At the point at which he makes this remark, he purports to have shown that the categories and concepts like reality and negation, substance and cause, express ways in which we must, as he puts it in the Prolegomena to Any Future Metaphysics, “spell out” [ buchstabieren ] the structure of perceived objects, lest we lose our grip on their being perceived objects: that is, genuinely perceived, rather than imagined [ Prolegomena 4:312]. He has argued that the categories are, moreover, no more than this: that their significance is exhausted by this role that they play in spelling out the structure of the perceptible. In saying this, he takes himself to disagree with those of his predecessors who believe that these concepts refer to something constituted independently of the human intellect, something to which, in exercising the intellect, we might hope to be attuned. Here is what he says about one version of that view, a version he calls a “preformation system”:

Note the parallels between this and heteronomist constitutivism. There is meant to be something that is there, independently of us—here, the “laws of nature,” there, the good—and we are predisposed by virtue of the constitution of our subject to think, in the theoretical case, or act, in the practical case, in accordance with it. In this passage, Kant says that this notion of accord, in the theoretical case, is a sham. The preformationist wants it to lie in our nature that we are connected to an order of causal intelligibility outside of ourselves, by virtue of a disposition that we have to think in a certain way, a way that matches up with that order of intelligibility. The trouble is that, on the preformationist’s picture, even if this is true, that this is an order of intelligibility is not reflected in our cognitive acts themselves. When I judge that the sun warms the stone, I do so not because the sun warms the stone (even if that is true), but because I am disposed to make such a judgment in my circumstances. Thus, as soon as the disposition is introduced to the explanatory story, the external order of intelligibility is bleached of its explanatory import in connection with my judgment. This is why Kant says, even under the assumption of that external order and a match-up between our disposition and it, that when I exercise this disposition in applying the concept of cause, I do not count as grasping the causal relation that is really there; I’m just chugging along as I’m disposed to.

Not every rationalist is a preformationist about the intellect, of course; here Kant has Crusius in particular in mind, as is explicit at Prolegomena 4:319n. A rational intuitionist, for example, would say not that the intellect has implanted dispositions to judge in such-and-such a way but that we have non-sensible modes of access to objects of knowledge. But the heteronomist constitutivist is analogous to the preformationist, and not to the rational intuitionist, in the sense I have indicated. This is true even of the heteronomist who embraces a genuine explanatory connection between the will and the good. This is precisely because, as I have emphasized, the will, unlike the theoretical intellect, is an appetitive or desiderative power: it is not a power to know anything prior to itself. This is what makes rational intuitionism unavailable in the practical case. This means that even a heteronomist who holds that the good is implicated in the nature of the will as a capacity must admit that in an exercise of the will, I do not know my exercise to be explained by the good.

This is not to say, of course, that the heteronomist must accept that we live in total ignorance of the good and its explanatory connection to acts of the will. On the contrary, she can, and typically does, say that we know the good and its relation to the will and the latter’s exercise in an exercise of the intellect, and even that such an exercise of the intellect can itself explain the exercise of the will.Footnote ¹⁸ But what Kant can say, which the heteronomist cannot, is that there is nothing to the good and what it is to pursue it that escapes the exercise of rational appetite itself: nothing to the good that escapes the self-consciousness of volitional activity.

To say this is not, of course, to offer an argument to the heteronomist with the ambition of immanent conviction. My aim has been to reach a more precise description of Kant’s disagreement with the heteronomist rather than a judgment in his favor. Kant’s alternative will, of course, be attractive to those who share his general position on a priori knowledge, to the effect that we cannot know a priori anything constituted independently of our powers of knowledge, especially if they also believe, as he does, that the good is not constituted by the nature of the theoretical intellect, since in that case the only thing left for it to be constituted by is the will. But even if we set aside Kant’s general position on the a priori, this section’s consideration of Kant’s arguments has added something to the previous section’s statement of his position: namely, that heteronomist constitutivism not only makes the good constitutively external to the will but also, because the will is an appetitive power, makes the good as such epistemically external to the will’s exercise. It may be that there are reasons apart from Kant’s general position on the a priori to object to the latter implication.Footnote ¹⁹

6. Objections and Replies

Before I conclude, I want to consider two instructive Kant-internal objections to the argument against heteronomist constitutivism that I have just presented. First, there is an objection to Kant’s anti-preformationist argument, which occurs to many readers of the first Critique, and an analogous objection arises in the practical case. Kant himself, like Hume, denies that in making causal judgments, I am attuned to a causal order constituted independently of my cognition. So has he not merely joined the empiricists against the rationalists? Likewise, Kant himself, like Hume, denies that in the exercise of the will, I am attuned to a normative order constituted independently of it. So must he not agree with Hume that I do what I do not because it is good but merely because I am disposed to do it?

No.Footnote ²⁰ For Kant, there is no notion of the good defined independently of the will. And this makes him fundamentally different from the sentimentalist. Sentimentalism concedes to rationalism that there is something which it would be for us to be connected to an external normative order; it just denies that we are thus connected. The very articulation of the sentimentalist position, as a skeptical moral-psychological position, depends on the availability of the notion of such a connection. That may seem dubious, inasmuch as Hume denies that we have ideas of right and wrong [Treatise III.i.i (Hume, Reference Hume1978: pp. 467–9)]. But we do, for Hume—at least as Kant reads him—have a sense of what the ideas of right and wrong would have to be if we had them. He says that though we thought our moral thought and talk were about an external normative order, they are not: actually, they’re just an expression of feeling. If there is a normative order in things, it is not one that we can know or have any idea of. This is parallel to his skeptical empiricism in the theoretical case. There he says, we can know no causal relations in things; causal judgments are just the expression of feelings (in that case, of expectation developed in reaction to constant conjunction). If there is a causal relation in things, it is not one that we can know or have any idea of. The articulation of these positions depends on a distinction between things, unstained by the palette of our minds, and such painterly embellishments as we bestow upon them. The rationalist and Hume disagree about how much we can justifiably say about what is in the things as opposed to what we project on them.Footnote ²¹

Kant’s anti-skeptical move, which is equally his anti-heteronomist move, is to deny the availability of the notion of things independent of us and our palette required to articulate both the dogmatic rationalist and the skeptical empiricist positions.Footnote ²² The skeptic can make us look parochial only if she can compare what we say and feel about things with what they, independently of that, really are; we can say to what extent our representations of the things conform to them and to what extent they do not. But Kant does not permit any such comparison, since for Kant, it is not our a priori representations that conform to things, but the things that conform to our a priori representations. His Copernican revolution, in the practical as in the theoretical case, is as antiskeptical as it is antidogmatic: for it is the rejection of the dogmatic assumption upon which skepticism depends just as much as dogmatism does, that in philosophy we may help ourselves to a general notion of things, the real, the world, and what’s out there, understood as a notion intelligible independently of any notion we may have of our own intellect and will.Footnote ²³

But there is a second objection, against which Kant’s Copernicanism may seem, at first, to be of no use at all, and that is that it leaves open that the constitution of our intellect and will may be ridiculous rather than sublime. In particular, it may seem that it leaves open that bisculinarism is a perfectly good description of our moral nature, alongside Kant’s preferred description in terms of the moral law. After all, the bisculinarist seems to count as an autonomist constitutivist, since she denies that the internal standard of the will is grounded in a normative order outside its nature.Footnote ²⁴ And if so, can we not after all generate a sense in which our nature is idiosyncratic, and thereby see Kantianism collapse into something like sentimentalism?

This is a topic that I cannot treat adequately here.Footnote ²⁵ Kant himself would want to exclude bisculinarism by arguing that it is unintelligible that anyone might be moved to pursue the gluttonous consumption of biscuits apart from happening to be inclined towards it; thus, he would say, bisculinarism does not offer an intelligible principle of fundamental end-selection. Only the moral law does. I myself am not sure whether he can explain why, independently of its being fully intelligible to us, constituted as we are, that someone might do something solely and fundamentally because she owes it to someone else, even if she does not want to do it; whereas it is not intelligible to us, constituted as we are, that something might do something solely and fundamentally because it would allow her to eat biscuits, even if she does not want to eat biscuits.

It may be that we can say more than this. It might be suggested that further conditions of adequacy on principles of fundamental end-selection rule out at least some competitors to the moral law: for example, that the principle be serviceable as that of nature, or even as that of a living nature.Footnote ²⁶ And certainly Kant seems to make such suggestions himself, and they may be correct. But even if they are not, I suspect that he can make good on his claim that the moral law is the only principle of fundamental end-selection whose intelligibility we are in a position to appreciate, and that this is enough to avoid a judgment of idiosyncrasy akin to the one characteristic of sentimentalism. This is because the mere logical possibility of bisculinarism (if it is logically possible) does not make it a really intelligible competitor to Kant’s conception of the will—that is, a real possibility that we can know and understand—any more than the mere logical possibility of forms of intuition different from our own make them really intelligible competitors to space and time.Footnote ²⁷ To judge something to be an idiosyncrasy, we must compare it with something else and see that it is different. But bisculinarism is not a something, not a reality, to which we can compare anything; it is, as far as we know, no more than a logically consistent description, in which we can understand no reality, nature, essence, or form, since a bisculinarist is, to us, completely unintelligible as a motivational and normative possibility.