Are our actions morally good because we approve of them, or are they good independently of our approval? Are we projecting moral values onto the world, or do we detect values that are already there? Rather than laying out genuine alternatives, these questions will be taken by many to articulate a secular variant of the Euthyphro dilemma: If our actions are good because we approve of them, moral goodness appears to be arbitrary. If they are good independently of our approval, it is unclear how we come to know their moral quality and how moral knowledge can be motivating. None of these options seems attractive; the source of moral goodness remains unclear.

Despite the growing literature on Kant’s moral epistemology, Kant’s answer to this apparent dilemma remains open to question. According to the moral realist, the objectivity of our moral judgments can only be preserved if moral values are independent of our voluntary approval. Objectivity requires that there be mind-independent moral facts so that our moral judgments can be true or false. Since Kant wants to hold on to the objectivity of our moral judgments, he would need to endorse the second horn of the dilemma. Thus, many have tried to ascribe some sort of realism to Kant. Allen Wood, for example, claims that “[s]ince Kant holds that moral truth is irreducible either to what people think or to the results of any verification procedures, he is a moral realist in the most agreed-upon sense that term has in contemporary metaphysics and metaethics” (Wood 1999, 157). According to the moral realist interpretation, the categorical imperative in its universal law formulation is merely a reliable goodness-tracking device. It detects moral goodness but is not itself the source of the good (Stern 2012; Stratton-Lake 2000; Langton 2007; Timmons 1998 et al.). The mind-independent moral facts that the moral realist views as necessary for objectivity are facts about how our actions relate to what is absolutely valuable: our “humanity”.

Contemporary Kantian constructivists hold that the procedure itself grounds the normativity of our value judgments. “Values for Kant are constructed by a procedure of making laws for ourselves […] [and are then] projected onto the world” (Korsgaard et al. 1996, 112). Many have argued that Rawls’s and Korsgaard’s versions of constructivism make moral facts dependent on our choices and thus commit them to anti-realism (Cohen 1996; Nagel 1996; Stern 2012). According to these critics, Kantians must give up on either the objectivity of moral judgments or constructivism. Korsgaard, by contrast, believes that even though she rejects the second horn of the dilemma, this does not necessarily lead to anti-realism or subjectivism. Since there is a correct procedure through which we arrive at our particular moral judgments, their objectivity is preserved, and moral values are not merely subjective.

Sharon Street has recently argued that Rawlsian constructivism is merely “restricted”; it cannot count as a metaethical theory at all, because it begs the question at issue. It purports to explain why and how something has value, but in fact the procedure (the original position and the veil of ignorance) simply presupposes the goodness of freedom and equality. All other results of the construction procedure owe their value properties to this initial presupposition. Korsgaard’s constructivism attempts to overcome this restrictedness and is, in contrast to Rawls’s constructivism, not merely a normative constructivism but indeed a metaethical constructivism. Korsgaard believes that her constructivism does in fact “go all the way down” (Korsgaard 2008, 324). According to Street, however, her attempt is based on the false assumption that we can derive certain “substantive values” from a “purely formal understanding of the nature of practical reason” (Street 2008, 244).

This rough sketch reveals both that there are alternatives to Kantian constructivism when it comes to making sense of Kant’s metaethics and the dissatisfying way in which some conceptual distinctions are drawn in contemporary metaethics. Thinking through Kant’s theory will put us in a position to see some of the presuppositions we have been taking for granted more clearly. The Kantian view I attempt to lay out in this paper aims to dissolve the secular version of the Euthyphro dilemma. In responding to this dilemma, we need to reach a better understanding of the source, or the origin, of our moral knowledge: Is it voluntary approval or mind-independent moral facts? Projectivism or detectivism? Constructed or given? I believe that each of these ways of articulating the problem presents a false alternative on closer inspection.

This paper consists of three parts. I will first show why what we nowadays call moral realism is not even a moral theory on Kant’s view. Moreover, I will argue that the realist attempt to reduce the universal law formulation of the categorical imperative to a mere goodness-tracking device fails. Hence moral realists cannot convincingly show that Kant endorses the second horn of the Euthyphro dilemma. In part two, I will turn to Rawls’s and Korsgaard’s versions of constructivism. Rawls’s account turns out to be an inconsistent marriage of realism and anti-realism. His weak notion of objectivity cannot account for unconditional obligation. Korsgaard’s argument not only turns Kant’s argument on its head but is also invalid. These positions are neither solutions to the Euthyphro dilemma nor genuinely Kantian. With this said, I still believe that there are important lessons to be learned from both realism and constructivism. In the final section, I would like to ascribe to Kant a moral idealism which can preserve the insights of both moral realism and constructivism without running into the same difficulties. I believe that the Euthyphro dilemma can be dissolved if we take seriously Kant’s idea that practical reason (the will) is a capacity for knowledge (Bagnoli 2012, 2013; Engstrom 2009).

1 Moral Realism

Moral realists believe that the strong sense of objectivity in Kant’s moral theory can only be secured if he embraces the second horn of the dilemma. They claim to find textual support in passages such as the following:

The essence of things is not altered by their external relations, and that which, abstracting from these, alone constitutes the absolute worth of man, is also that by which he must be judged, whoever the judge may be, and even by the Supreme Being. (GMM IV, 439)

Allen Wood takes this passage to be a clear indication that Kant is not an anti-realist or, what amounts for him to the same thing, a constructivist. He reads this passage as follows:

Human beings have absolute worth, which belongs to them essentially. This worth is not something conferred on them by themselves, or by God, or by anybody else. No being’s stances, attitudes, judgments, or “legislative acts of will” are required for rational beings to have that worth, because they have it essentially—and that is the sole and sufficient reason why everyone, even God, should judge them to have it. (Wood 2008, 112)

Wood’s main inference is that if moral value belongs to human beings essentially, then their moral value cannot depend on their own, or someone else’s, attitude. Kant is a realist because he “holds that moral truth is irreducible either to what people think or to the results of any verification procedures” (Wood 1999, 157). Wood’s moral realist interpretation is certainly correct in its rejection of any voluntaristic misunderstanding of Kant’s notion of autonomy. The fact that autonomy, the capacity to give a law to ourselves, is the highest principle of Kant’s ethics does not imply that we arbitrarily decide what counts as good or evil. Kant is neither a voluntarist nor a subjectivist. The nomos, the law, that we give to ourselves is not an individual invention, but is constitutive of rational agency as such. I take this to be the common ground between realists and constructivists. Kantian realists reject constructivism, however, because they believe that it leads to subjectivism. They infer from the fact that Kant is not a subjectivist that he must endorse the second horn of the Euthyphro dilemma. I believe that we may resist this inference. But for now, I will go along with the Kantian realist and reserve the constructivist response for the second section.

Moral realists can argue that if Kant’s answer to the Euthyphro dilemma were not the realist one, one would expect moral realism to show up as an alternative account on his table of all misguided moral theories. In this table, Kant distinguishes between moral theories that are based on an empirical determining ground and those that claim to have their determining ground in reason (GMM IV 441 f., CPrR, V 40). According to Kant, all moral theories in the history of moral philosophy can be subsumed under one of the two sides of the table. However, as it turns out, none of these moral theories has been able properly to account for moral obligation. Kant claims that this is because all theories other than his own are based on the same false principle: the principle of heteronomy. Even those theories that claim to be based on reason are revealed, on closer inspection, to be based on empirical grounds. With the principle of autonomy, Kant believes that he has found, for the first time in the history of philosophy, the only true principle of morality. For only if our capacity of volition in its legislative and executive function depends on pure reason rather than some presupposed desire is its claim to unconditional obligation justified. Since only a moral theory based on the principle of autonomy has its determining ground in reason, Kant implicitly also holds that his ethics is a more consistent rationalist moral theory compared to the traditional accounts.

As we will see, none of the positions in Kant’s table is fully in line with moral realism. Thus moral constructivists face the following dilemma: Either they must concede that Kant did not show why moral realism is based on a heteronomous principle (and thus that he has not ruled out the ability of other moral theories to account for unconditional obligation) or they must admit that Kant is in fact a moral realist, and that he did in fact endorse the second horn of the secular version of the Euthyphro dilemma. This would also imply that moral constructivism is either not fundamentally Kantian or not an alternative to moral realism.

In this section, I want to claim that the reason why moral realism falls outside the table of all misguided moral theories is that it does not even qualify as a moral theory; Kantianism in metaethics should not be equated with moral realism. Moreover, I would like to show, on textual grounds, that Kant explicitly rejected the possibility of moral realism. So I think constructivists are right when they claim that Kant’s position is an alternative to both moral realism and anti-realism. As I will argue, however, this does not validate the inference that Kant was a moral constructivist. Kant’s notion of autonomy sets him apart from both moral realism and constructivism.

Let’s take a closer look at Kant’s table of heteronomous moral theories. One might argue that moral realism is in principle identical to a heteronomous position that Kant subsumes under moral rationalism, i.e. “outer perfectionism”. Outer perfectionism is the view that our moral commands are based on the concept of God’s perfection. One might think that if moral obligation is dependent on the concept of God’s perfection, it is just as mind-independent as moral realism. Moreover, since it also seems to decouple moral values from our volition, it seems to run into the same problem of moral motivation. If this were true, moral realism would, in the end, be closer to outer perfectionism than to Kant’s moral theory. Moral realists would be perfectionists rather than Kantians.

However, Kant’s argument against outer perfectionism makes clear why outer perfectionism is fundamentally distinct from moral realism. Kant first claims that “perfection” needs to be understood in a practical rather than a metaphysical or transcendental sense. Kant’s idea here seems to be that only in its practical sense can “perfection” function as a concept of a moral theory. I’ll come back to this point in a moment. “Outer perfection” in a practical sense is “suitability of God to all ends in general”. This principle, according to Kant, is empty because it cannot by itself determine our ends but rather presupposes some ends as given. Since these ends are presupposed, the determination of our volition through the concept of perfection is ultimately not based on a law we give to ourselves. Hence perfectionism is not based on the principle of autonomy. Moreover, since these ends are given to us from elsewhere, perfectionism collapses into subjectivism or, as Kant would put it, into “empiricism”. In other words, none of the moral theories in the table can escape empiricism. Kant’s Critique of Practical Reason is not a Critique of Pure Practical Reason, because, in its practical use, it is only the empirical rather than the pure employment of reason that needs to be criticized and restricted (CPrR V, 3). On Kant’s view, his moral philosophy is the only one that has its source in pure reason. All other moral theories turn out to be empiricist. Thus they cannot account for unconditional obligation and are fundamentally misguided.

It should be clear at this point that Kant’s heteronomy objection does not call into question how, on the traditional accounts, moral commands can have a motivational grip on the agent. On the contemporary picture, the concept of autonomy is supposed to close the motivational gap between our mere knowledge of what is good and our doing what is good. Kant’s point, by contrast, is not merely that autonomy can explain how it is that we are motivated to x. His point is rather that the claim to unconditional necessity, which he takes to be a necessary implication of our moral judgments, would be unjustified if our capacity of volition were merely heteronomous.

If moral realism were equivalent to outer perfectionism, the heteronomy charge would equally apply to it, and its claim to unconditional necessity would be unjustified. What makes the issue more complicated, however, is that moral realism is not identical to outer perfectionism. If we are determined, as the realist believes, by our cognition of some metaphysical moral fact of which we have intuitive knowledge, our will must be determined not by some presupposed end but by the cognition of this metaphysical moral fact, and hence by a cognition of what is objectively good or bad. Thus, moral realism is not open to Kant’s heteronomy objection, and moral realists have rightly denied that they endorse practical perfectionism. As Robert Stern points out:

While Kant offers a critique of perfectionism for its heteronomy, he does not offer a critique of rational intuitionism as such on similar grounds. […] It is no accident that Kant offers no such critique of rational intuitionism: for, given the distinction between autonomy and heteronomy that Kant is working with here, any critique of this sort would be out of place. (Stern 2012, 25 f.)

Autonomy and heteronomy are concepts that refer to our faculty of volition. Anti-realists cannot consistently run the heteronomy objection and the argument from motivation against moral realism at the same time. Were Kantian moral realists open to the heteronomy objection—were they to presuppose some given end—there would be no sense in which they decouple moral cognition and volition. And to the extent that they don’t decouple moral cognition and volition, they don’t run into the problem of motivation.

Now, one might think that since moral realism can escape the heteronomy charge, the realist solution to the Euthyphro problem is also Kant’s solution, and what remains to be solved for both the moral realist and Kant is the problem of motivation. Unfortunately, the matter is not quite this simple. As I mentioned above, Kant only considers the practical notion of “perfection”. A “transcendental” or “metaphysical” notion, which does not have any relation to our faculty of volition, is not even taken into consideration, because it could not be part of Kant’s table of “possible principles of morals from the fundamental concept of heteronomy” (GMM IV, 441). Kant only considers the concepts of autonomy and heteronomy as fundamental concepts of moral philosophy because he believes that moral philosophy must necessarily relate to our faculty of volition if cognition is to be practical. This is bad news for Kantian moral realists; moral realism is excluded from the table of heteronomous moral theories not because it is based on the same principle as Kant’s moral philosophy, the principle of autonomy, but because it does not qualify as practical philosophy at all. In responding to the heteronomy objection, the moral realist throws the baby out with the bathwater. By decoupling moral values from volition, moral realists not only face the problem of moral motivation but must also explain how their theory can count as an account of practical cognition at all. In responding to the heteronomy objection, moral realists must admit that moral values merely stand in a theoretical relation to us. It therefore remains unclear how moral cognition itself can be practical, which is precisely the core claim of Kant’s moral philosophy (CPrR V, 31).

The realist may respond that there is no reason to assume that our moral beliefs cannot also be motivating; it is dogmatic to assume that our desires need to exist independently of our moral beliefs. When we judge x to be good, we are also motivated to x because we deem x to be good. Kantian moral realists may add that respect for the moral law in Kant stands in the same relation to our moral beliefs. It is not some separate moral desire, but the volitional (or conative) aspect of the cognition of moral facts. However, with this response moral realists either fall back into heteronomy or abandon moral realism. If moral values exist independently of our volition, they must affect us such that we are motivated to x. If our actions were determined by sensible affection, there would be no room for “acting from duty”—at best, there would be room for “acting in accordance with duty”. Hence with this response we fall back into heteronomy. To the degree that moral realists hold that the cognition itself brings about some motivational state, i.e. that our belief that x is good is identical to being motivated to x, they abandon the view that moral values are mind-independent. Taking this tack therefore involves giving up on moral realism. If Kantian moral realists were to claim that our belief that x is good and our motivation to x necessarily come together but stand in a merely accidental relation to one another, acting from duty would be impossible. To hold this view is to give up on Kantianism.

Kantian moral realists encounter even more fundamental difficulties when they attempt to explain what it is about moral values, according to Kant, that motivates us. Since they are committed to the claim that moral values exist independently of the mind, they cannot believe that the “CI procedure”, i.e. the act of rational volition, is constitutive of moral values. Instead, they consider the “procedure” to be a goodness-tracking device. The universal law formulation of the categorical imperative merely plays a “criterial role”. It functions as a “test [of] our moral judgement, but is not itself a normative reason why certain acts ought to be done” (Stratton-Lake 2000, 77). What this device tracks, according to realists, is the core moral value: “our status as free and rational agents”, i.e. our humanity (Stern 2012, 27; cf. Langton 2007, 182; Guyer 2000, 147). But what is it about rational agency that makes it valuable? Why are our inclinations of merely relative value, whereas our rational agency is of absolute value? According to Kantian moral realists, Kant simply assumes the “axiological priority” of the noumenal over the sensible world (Schönecker 1999). As rational beings we belong to the noumenal world, and hence we have absolute value. Yet Kant does not have an argument that demonstrates the priority of the noumenal over the sensible world. Nor can he explain why humans ultimately identify more with reason than with their sensible inclinations (Stern 2012, 31). Thus Kant cannot give us a satisfactory account of why morality is motivating (Schönecker 1999; Stern 2012). Moreover, and most fundamentally, in reducing the universal law formula to a mere goodness-tracking device, moral realists misconstrue the connection between our practical rationality and the value of our humanity. The essence of human beings, their humanity, consists in being “ends in themselves”. “Being an end in itself” signifies the human being’s capacity to act from the representation of laws. The universal law formula demands precisely this: exercise your capacity, become what you fundamentally are—a being who acts from the representation of the universality of its principles (i.e. acts from laws). The moral realist interpretation tends to ignore or downplay the systematic conception of both formulas. On the realist account, the value of our humanity must be thought of as a value property that exists independently of practical cognition. In Kant, however, to be a practical cognizer is to act from the consciousness of universal principles, which is the consciousness articulated in the universal law formula. In downgrading the universal law formula to a mere goodness-tracking device, the realist separates what for Kant essentially belongs together: universal volition and absolute value.

To sum up: Kantian moral realists cannot convincingly establish that Kant did in fact endorse the second horn of the Euthyphro dilemma. Their conception of the categorical imperative as a goodness-tracking device is fundamentally mistaken. They detach moral value from our capacity of volition, and in doing so they not only fail to account for how moral values become motivating but also abandon the very idea of genuine practical cognition.

2 Moral Constructivism

2.1 Normative Constructivism

Constructivists run into the Euthyphro problem from the other end, as it were. They have to explain how they can account for objectivity. Since constructivists make moral goodness dependent on our volition, they are considered subjectivists (Cohen 1996; Nagel 1996; Stern 2012). It is hard to see how subjectivism is compatible with objectivity in the Kantian sense. On the subjectivist account, what we deem to be good is not that our volition is practically universal, but some merely subjective end given to us from elsewhere. Hence we can never act from the representation of a practical law (i.e. a universal practical principle) that we give to ourselves, and thus we fail to be autonomous. Voluntarists might say that the principle they deem to be good is based on free choice. But if what we choose is merely accidentally related to us, because it could have been otherwise, it is hard to see how there could be any nomological agency, i.e. autonomy and objectivity, at all. So even if subjectivism can explain why we are motivated to do something, it cannot explain how unconditional or objective moral volition is possible.

John Rawls responds to the Euthyphro dilemma by embracing the first horn and officially rejecting what he calls moral intuitionism. Rawls defines moral intuitionism as the view according to which there is a “moral order that is prior to and independent of our conception of the person and the social role of morality. This order is given by the nature of things and is known […] by rational intuition” (Rawls 1980, 557). This definition makes it clear that Rawls does not consider intuitionism to be a merely epistemological doctrine. His definition contains both an epistemological and an ontological aspect: The moral order is known by rational intuition, and it exists independently of our conceptions. I think it is fair to say that, in objecting to moral intuitionism, Rawls also objects to moral realism, or at least to the type of moral realism that holds that moral properties exist mind-independently. (For the sake of consistency, I will, in the remainder of this paper, use the term “realism” rather than Rawls’s preferred “moral intuitionism”.)

The puzzle for Rawls is how he can reject the realist horn of the Euthyphro dilemma without thereby falling back into the subjectivism of the first horn. Rawls believes that constructivism is a solution to this puzzle. Constructivists can embrace the first horn in such a way that they set themselves apart from both realists and subjectivists. At the base of Rawls’s argument lies a different concept of objectivity: “[T]he rational intuitionist notion of objectivity is unnecessary for objectivity” (570). Objectivity in moral theory, Rawls claims, ought to be “understood by reference to a suitably constructed social point of view”:

Kantian constructivism holds that moral objectivity is to be understood in terms of a suitably constructed social point of view that all can accept. Apart from the procedure of constructing the principles of justice, there are no moral facts. (Rawls 1980, 519)

With this account, Rawls wants to secure a strong notion of moral objectivity without embracing moral realism. I am not going to assess Rawls’s arguments against moral realism here. There are good reasons to believe that they are at least insufficient to disprove it (cf. Stern 2012). The question that interests me is how Rawls’s constructivism accounts for moral objectivity and whether this account can accommodate the kind of moral obligation Kant is after. I am going to show that Rawls fails to deliver a genuine alternative to the realist and subjectivist answers to the Euthyphro dilemma. Moreover, his own answer to the dilemma is fundamentally incompatible with Kant’s ethics.

In the above passage, Rawls holds that moral objectivity implies possible universal agreement. A moral judgment is objectively valid if “all can accept” it. This is what Kant calls “subjective universality”. Yet, in contrast to Kant, the agreement is based on “a suitably constructed social point of view”. So what is to be constructed is not only our particular moral judgments, but the “social point of view” in general. Only after we have constructed the “social point of view” can we derive particular judgments from it. If these derived particular moral judgments are in line with our considered moral judgments, we have good reason to believe that the social point of view is in fact “suitably” constructed. On Rawls’s account, the original position generates the two most fundamental principles of justice, which in turn determine our particular moral judgments. It is important to note that construction is not only a top-down activity. The fundamental normative judgments, the two principles of justice, are also constructed (just like the resultant particular moral judgments). If the resultant particular moral judgments turn out to be unreasonable, the two principles of justice have to be corrected. However, even if we allow for construction to be a top-down and bottom-up procedure, the construction has to end somewhere, for it is unclear how the constitutive standards of the construction procedure can be constructed without running into vicious circularity. If the activity of construction is guided by normative principles, these principles cannot themselves be the result of construction. They must in some sense (not necessarily in the realist sense) already be there. For this reason, Rawls’s ontological claim in the above passage is puzzling. If there are “no moral facts […] apart from the procedure of constructing the principles of justice”, how does the procedure get its normative content (Rawls 1980, 519)? There has to be some constitutive or normative principle that is not constructed and is normative for the activity of construction itself.

Part of the problem is that the concept of “construction” is notoriously unclear. We can distinguish between at least three different aspects. The activity of construction is guided by a procedure, which in Rawls’s case is the procedure of “justice as fairness”. The result of this procedure is particular moral judgments. Finally, construction requires material. In Rawls’s case, this material is the freedom and equality of moral persons and society as a fair system of cooperation. Even if our particular moral judgments can be derived from this material together with the construction procedure, the procedure itself, as well as the initial material input, cannot be constructed.

One possible reply to this is to give up on the idea that construction, at its most fundamental level, is a rule-governed activity. What Rawls could mean is that we make a radical choice about what the construction procedure should be like. Once the rules are fixed, we arrive at our resultant moral judgments and see whether they cohere with our considered moral judgments. With this move, however, the moral constructivist throws the baby out with the bathwater. If constructivism is at bottom some kind of voluntarism, the universal agreement Rawls attempts to secure is merely accidental or at best, as Kant would put it, physically rather than practically necessary (CPrR, V 26). If Rawls were to embed the standards of the procedure in our choices, constructivism would fall back into subjectivism and anti-realism. Our moral judgments would be at best intersubjectively but not objectively valid. As a kind of anti-realism, Rawls’s constructivism would fail not only to offer a genuine alternative to other accounts in metaethics but also to account for the kind of objectivity a truly Kantian moral theory must embrace.

However, in Kantian Constructivism, Rawls anticipates the voluntarism charge. Against constructivism, he writes:

The rational intuitionist may object that […] first […] principles are not the kind of thing concerning which it makes sense to say that their status depends on their being chosen or adopted. We cannot “choose” them; what we can do is choose whether to follow them in our actions or to be guided by them in reasoning, just as we can choose whether to honor our duties, but not what our duties are. (Rawls 1980, 567)

Rawls does not want to say that the parties in the original position make a “radical choice” or an existentialist choice: a choice “not based on reasons […] that simply fixes, by sheer fiat […] the scheme of reasons that we […] are to recognize”. Instead, our choice in the original position is “subject to constraints that express reasonable conditions” (Rawls 1980, 568). These constraints include the “condition of publicity”, “the veil of ignorance”, and the “symmetry of the parties’ situation” (Rawls 1980, 530).

As much as this response suffices to fend off accusations of voluntarism or existentialism, it still leaves unexplained how we come to know the “constraints that express reasonable conditions” and what the ontological status of these conditions or reasons is; it leaves unexplained what makes these constraints authoritative or what makes them reasonable, i.e. something we are rationally compelled to accept. Rawls claims that these constraints best express our nature as free and equal citizens. But why ought we to take freedom and equality to be of fundamental value? Are freedom and equality values that exist mind-independently, and do we come to know them through intellectual intuition? Rawls does not want to commit himself to anything as strong as this. His modest answer is that he does not want to settle the question of justice “regardless of [the] particular social or historical circumstances. […] We look to ourselves and to our future, and reflect upon our disputes since, let’s say, the Declaration of Independence” (Rawls 1980, 518). This approach can hardly count as a Kantian approach to morality—not even, as Rawls believes, in an “analogous” sense (Rawls 1980, 517). First, Rawls seems to give up on Kant’s idea of a historically invariant moral principle. Instead, he is willing to limit the scope of objectivity to a particular time and place in human history. Since this notion of objectivity has a time index, it is hard to see how Rawlsian objectivity can possibly establish the unconditional demands that are so crucial for Kant. Freedom and equality need to be presupposed in moral deliberation, but not merely because they are fundamental values of a liberal democracy. On Kant’s account, freedom just is the capacity to act from universal principles, i.e. to value unconditionally. Only in virtue of this capacity do we have absolute moral value, i.e. dignity, and only with respect to it are we fundamentally equal. But this capacity to act from universal principles does not exist mind-independently, since it is the capacity of practical reason itself (GMM IV, 412).

Rawls holds that “there are no moral facts […] apart from the procedure of constructing the principles of justice” (Rawls 1980, 519). He does not want to commit himself to moral realism. But his procedure seems to be grounded on two moral facts, namely the value of freedom and the value of equality, which are not constructed through the procedure but instead function as the initial construction material. Since Rawls seems to take these moral facts as given, his position collapses into moral realism (Stern 2012, 27; Langton 2007, 182; Guyer 2000, 155). Thus what Nadeem Hussain and Nishi Shah have pointed out with respect to Korsgaard applies equally to Rawls: “[S]ome […] metaethical position needs to be added to constructivism in order to turn it from an account of which normative judgments to make into an account of what it is to make a normative judgment” (Hussain and Shah 2006, 292).

One might argue that Rawls did not really attempt to establish a metaethical constructivism and instead intended to provide a merely normative or “restricted” constructivism (Street 2008, 209). This kind of constructivism would not need to go all the way down and would therefore allow him to be a realist about the fundamental values. If this were correct, we would be mistaken to expect Rawls’s constructivism to solve the Euthyphro dilemma. And in truth, I believe that holding on to Rawls’s normative account is the best we can do for his constructivism. However, the fact that Rawls explicitly presents his constructivism as an alternative to moral intuitionism, which he clearly defines in metaethical terms (i.e. non-reductivism, self-evidence, mind-independency [Rawls 1980, 557]) suggests that he intended his constructivism to serve as an explanation of the ultimate ground of goodness.

Moreover, by weakening the notion of moral objectivity, Rawls throws the baby out with the bathwater. The weaker notion of objectivity cannot give us the Kantian kind of objectivity needed for unconditional moral demands. If Rawls were to reconcile constructivism with moral realism by presupposing freedom and equality as moral facts, constructivism’s status as a metaethical theory (let alone a self-standing metaethical theory) would come under question, since it would be unable to explain the ultimate ground of our normative judgments. Understood in this way, it would seem only to provide arguments for how to get from one specific set of normative commitments to others.

2.2 Metaethical Constructivism

Christine Korsgaard’s argument from valuing our humanity can be seen as an attempt to deliver an argument where Rawls’s constructivism falls short. She explicitly holds that constructivism can go all the way down. Her criticism is directed at moral realism, but it also applies to Rawls to the extent that his constructivism either does not go deep enough or collapses into moral realism. The moral realist, according to Korsgaard, tries to answer the normative question by stipulating that some things just have intrinsic normative properties. Since this claim begs the question, according to Korsgaard, classical moral realism fails. I shall not here discuss whether Korsgaard does justice to moral realism. Rather, I want to assess whether she can improve on Rawls’s constructivism and, if so, whether she does so on Kantian grounds. Following Rawls, Korsgaard turns to the construction procedure in order to get to the sources of normativity:

When an impulse—say a desire—presents itself to us, we ask whether it could be a reason. We answer that question by seeing whether the maxim of acting on it can be willed as a law by a being with the identity in question. If it can be willed as a law it is a reason, for it has an intrinsically normative structure. If it cannot be willed as a law, we must reject it, and in that case we get obligation. (Korsgaard et al. 1996, 113)

Korsgaard calls this test of our principles of action the “reflective endorsement test”. The reflective endorsement test is the construction procedure. The “pleasures and pains” are the material of construction “with which reason works in constructing its notions of what is best” (Korsgaard 2008, 171). The “maxims” or the particular moral judgments are the result of construction.

When we come to make a decision, we consider our reasons for acting; in choosing one action over another, we resolve the conflict within us, thereby restoring unity. Or, in other words, by “giving a law to ourselves” we determine what kind of person we want to be and thereby commit ourselves to acting in certain ways. However, since the value of acting in one way or another ultimately seems to depend on our individual choices, many have objected that Korsgaard’s constructivism, like Rawls’s, amounts to voluntarism. If this is correct, constructivism can hardly be seen as a middle position between realism and anti-realism; its solution to the Euthyphro dilemma becomes problematic, and the reason why this solution is supposed to be Kantian becomes unclear. Kant is clearly not a voluntarist. Whether I break my promise is up to me. Whether I am obligated to keep a promise is not up to me. In adopting a maxim, I decide whether I want to act on a good maxim, but I do not decide whether the maxim is good. Even if we interpret Kant’s claim that the moral law is a Faktum of pure reason in a literal sense, as some sort of activity, it is not as if we invent the law. What remains relevant is not the activity of one particular agent but that of practical reason in general. To put it rather poetically: In practical reason, we are one.

Korsgaard responds to the voluntarism charge, like Rawls, by pointing to the rational constraints that are constitutive of the reflective endorsement test. The procedure does not deem permissible every maxim that happens to be consistent with or contribute to being the kind of person we want to be. Instead, a maxim can only count as good “if action and […] purpose are related to one another so that the maxim can be willed as a law. It is the relation of the action and the purpose which determines the maxim’s moral quality”. The procedure therefore deems certain reasons to be objectively justified; they are reasons, and their claim to normativity is valid (Korsgaard et al. 1996, 108).

Korsgaard does not tell us much more about the procedure of reflective endorsement, i.e. how exactly the constraints are supposed to preserve moral objectivity. Rather surprisingly, she believes that the reflective endorsement test still leaves room for relativism. She seems to think that the normative constraints only come into the picture once we have adopted a particular personal identity. Yet what kind of human being we want to be or which personal identity we want to adopt—e.g. a murderer or a doctor—seems at that point to remain entirely up to us. Both Korsgaard and Rawls downplay the fact that in Kant the so-called “CI procedure” articulates practical knowledge. How can we know that something is a “practical law” in Kant’s sense and still doubt that we are obligated to follow it? Rawls, in line with his coherentist approach, thinks that the principles generated by the CI procedure merely “match more accurately than other views our considered convictions” (Rawls 1980, 568). Korsgaard introduces a reflective endorsement test, but she underestimates the status of its results. She believes that an additional argument is needed to account for the source of moral obligation: the argument from valuing humanity. “Humanity” is our “power to reflect on our impulses”. Possession of this capacity is a necessary condition for being a rational agent. If we stopped employing this capacity in action, we would cease to be rational agents. Yet we do employ this capacity in action. Hence, Korsgaard concludes, we do in fact value our humanity. Since we do not differ with respect to our humanity, we must value other people’s humanity as well (Korsgaard et al. 1996, 123). The upshot of this argument is that not all of our personal identities are up to us; our moral identity is necessary because it is constitutive of rational agency (ibid).

This argument is neither valid nor Kantian. Firstly, the argument is a merely conditional one. Insofar as we want to be agents, we need to value our humanity. But why would we want to be agents in the first place? Korsgaard fails to provide an argument for what it is about our agency that we find valuable (cf. Enoch 2006). Secondly, and more importantly for our concerns here, it remains unclear how this argument in any way involves the activity of “construction”, which gives rise to the label “constructivism”. It may be appropriate to describe the “reflective endorsement test”, which involves maxim formation, as an act of construction, yet the same does not hold for the way in which Korsgaard establishes the value of our humanity. Granted that valuing our humanity is constitutive of rational agency, as Korsgaard claims, it does not follow that its value is a normative fact established through construction. Even if we take “construction” in a very wide sense, as bringing something into existence in accordance with a rule, the value of our humanity cannot count as constructed.

Korsgaard could respond by saying that we take an interest in valuing our humanity because we want to be agents. But this reply puts her yet again on the voluntarist track, which she attempts to avoid and which is incompatible with Kant. Thus even the wide sense of “construction” does not succeed in bringing constructivists closer to Kant, nor does it make their project philosophically sounder. If Korsgaard’s argument is that we cannot help but value our humanity, as she also suggests, then again the necessity seems to be physical rather than practical.

To be sure, the Kantian sense of “construction” is not available to so-called Kantian constructivists anyway. For Kant, intellectual construction involves a priori intuition (CPR, B 741). Constructing a triangle means “exhibiting an object corresponding to this concept, either through mere imagination, in pure intuition, or on paper, in empirical intuition, but in both cases completely a priori, without having had to borrow the pattern for it from any experience” (CPR, B 741 f.). What is crucial for our concerns here is that for Kant cognition through construction is based on pure intuition a priori. The constructed object need not exist empirically but can exist merely as an ens imaginarium, an object of our imagination. It does not, however, exist as an object of mere thought. In the second Critique, Kant explicitly holds that the moral law is “a synthetic proposition a priori, which is neither based on pure nor on empirical intuition” (CPrR, V 31, my emphasis). Although this is one of Kant’s most mysterious formulations, this much is clear: The moral law as a “fact of pure reason” is a fact of reason, not of (pure) intuition. There is therefore no good philological reason to believe that fundamental moral cognition in Kant is cognition through construction. The crucial point here is that it is not an interpretive option for constructivists to view Kantian moral cognition as a case of construction (on his own understanding of the term). There is an important difference between theoretical and practical cognition, but this difference does not consist in practical cognition’s being cognition through construction (CPR, IX f., CPrR, V 46).

Finally, it is worth noting that Korsgaard’s argument appears to turn Kant’s argument on its head. Kant does not first attempt to prove the value of our humanity and then derive moral obligation from it. His claim is not that our humanity consists in our capacity to critically reflect on our impulses, and that this is the basis of our unconditional value. Instead, Kant’s claim is that our humanity consists in our capacity to act from pure practical reason, i.e. from the representation of universal (unconditional) principles. In other words, it is not because we must value our humanity that we are obligated to act from the representation of laws, as Korsgaard holds. Rather, as I will argue in the following section, we have (unconditional) value (“dignity”, and not merely a “price”) because we have the capacity to act from the representation of laws. Our capacity to act from the cognition of universal principles is the ground for the attribution of value to our humanity. Our humanity consists not merely in the power to “reflect on our impulses” but in our capacity to be an end in ourselves (GMS IV, 430), i.e. to act from the representation of laws (GMS IV, 412).

Korsgaard’s and Rawls’s attempts to make moral values constitutive of the procedure or, as Korsgaard puts it, of the “test of reflective endorsement”, are philosophically sound and fundamentally in line with Kant’s project. However, both Rawls’s and Korsgaard’s versions of constructivism fail adequately to respond to the Euthyphro problem. Rawls uses a watered-down notion of objectivity that fails to account for the unconditional necessity that is essential to Kantian ethics. Moreover, Rawls is in principle willing to think of the fundamental constraints on his procedure, “freedom and equality”, as moral facts to be intuited in the classic moral intuitionist sense. He thereby blurs the distinction between his constructivism and classical moral realism on a fundamental level. Korsgaard, in her attempt to justify moral obligation, turns Kant’s argument on its head. She tries to establish moral obligation through an analysis of the concept of rational agency instead of deriving the moral value of humanity from the cognition of moral obligation. Instead of solving the Euthyphro dilemma, Rawls’s and Korsgaard’s positions encounter another dilemma, which we might call the dilemma of constructivism: If moral facts are constructed, then they are not objective. If they are objective in virtue of a construction procedure, then this procedure cannot itself be constructed (see Hussain and Shah 2006, 291). Since constructivism does not want to give up on a strong sense of objectivity, it collapses into realism. I believe that Rawls and Korsgaard run into this dilemma because they do not fully commit themselves to Kant’s project. They fail to see practical cognition as the source of moral value. In the remaining section, I want to lay out Kant’s answer to the Euthyphro dilemma as I see it.

3 Moral Idealism

In order to bring Kant’s solution to the Euthyphro dilemma properly into view, it is essential to reflect on his conception of practical reason; this will reveal why the secular version of the Euthyphro dilemma is a false alternative. Korsgaard believes that the substantive moral realist conflates ethics with epistemology (Korsgaard et al. 1996, 44). This distinction is misleading, however. For Kant, reason is a single capacity with two employments—one theoretical, the other practical. In both its employments, reason is a capacity for knowledge. There is no inconsistency in claiming that Kant’s Groundwork and the second Critique constitute Kant’s practical epistemology. Both Rawls and Korsgaard downplay the fact that in Kant the so-called “CI procedure” articulates practical knowledge (Erkenntnis). Rawls believes that Henry Sidgwick and rational intuitionism more generally misconceive the problem of moral goodness as an “epistemological problem”. Korsgaard explicitly equates “epistemological” with “theoretical” (Korsgaard et al. 1996, 44). Rawls, in line with his coherentist approach, thinks that the principles generated by the CI procedure merely “match more accurately than other views our considered convictions”. Korsgaard introduces a reflective endorsement test, but she still thinks that the test leaves room for moral relativism. However, since (ex hypothesi) we know that the maxim cannot be willed as a universal law, there is simply no world (of rational cognizers) in which it could possibly qualify as a good maxim, hence there is no room for relativist doubts. Constructivists are therefore right to consider “the procedure” or, as one may also put it, rational volition (i.e. practical reason) as the source of moral values. Yet Rawls and Korsgaard haven’t fully fleshed out Kant’s idea that the “procedure” articulates synthetic a priori practical knowledge, or that the categorical imperative is a “synthetic proposition a priori”. This is the reason why constructivists cannot in the end satisfactorily respond to the voluntarism charge. Rawls’s and Korsgaard’s versions of constructivism cannot, as we have seen, go all the way down. Rawls attempts to overcome these difficulties by introducing a weak notion of objectivity. Korsgaard seeks to escape the subjectivism of the first horn with her question-begging argument from humanity. Both ultimately fail to establish a way out of the dilemma.

There is something fundamentally right about Korsgaard’s objection that the moral realist mistakenly “thinks of practical philosophy as an essentially theoretical subject”. Yet the realist’s mistake is not that she believes that practical philosophy seeks “to find, or anyway to argue that we can find, some sort of ethical knowledge that we can apply in action”. Korsgaard overlooks that it is indeed Kant’s project to establish the categorical imperative as the form of, to use Korsgaard’s phrase, “ethical knowledge”. Moreover, Kant wants to show how this formal principle lies behind all our particular moral judgments. Since these particular moral judgments are conclusions of practical syllogisms, which guide our moral action, Kant does in a sense explain how we “apply” the general ethical knowledge we have—the categorical imperative—“in action”.

The real reason why moral realism is not in line with Kant’s project is that it views practical reason as a theoretical capacity (Korsgaard 2008; Engstrom 2013) that cognizes values which exist independently of the capacity of practical reason. This turns moral cognition into a kind of representational or theoretical cognition. The moral law is reduced to a good-tracking device rather than the source of goodness itself, i.e. the form of the good (Stern 2012; Langton 2007; Stratton-Lake 2000; Timmons 1998). The realist holds that the underlying value, the mind-independent metaphysical fact, is the absolute value of human beings (humanity). The moral realist might be able to account for moral objectivity, but his answer to the Euthyphro problem is neither satisfactory nor Kantian. It fails to take into account the equivalence of the universal law formula and the formula of humanity. The essence of human beings, their humanity, consists in being “ends in themselves”. “Being an end in itself” signifies the human being’s capacity to act from the representation of laws, because only an end in itself can make the execution of its desires dependent on the form of practical volition (i.e. the idea of practical universality). The universal law formula demands precisely this: Exercise your capacity of pure practical reason; become what you fundamentally are—a being who acts from the representation of the universality of her principles (who acts from laws). The moral realist interpretation tends to ignore or downplay the systematic connection between both formulas. On the realist account, the value of our humanity must be thought of as a value property that we become aware of through intellectual intuition. Not only is this claim dogmatic, but it also directly contradicts Kant’s discursivity thesis. According to the alternative view, we only become aware of the absolute value of our humanity through the universalizability requirement, which is constitutive of a capacity for rational volition. There is no knowledge of the value of our humanity prior to or independently of our knowledge of ourselves as beings that can act from the representations of the universalizability of our maxims. In downgrading the universal law formula to a mere good-tracking device, the realist fundamentally misconstrues the connection between our rationality and the value of our humanity. Kant’s ontology preserves a very strong sense of objectivity without thereby giving in to the idea that moral facts exist independently of us.

The moral law as the form of practical cognition is, on Kant’s account, “self-consciousness of practical reason” (CPrR, V 29); there is no object that is ontologically prior to this cognition. In practical cognition, practical reason cognizes itself, i.e. its own requirements of rational volition. Thus we do not first presuppose some value property of our rational nature as an end in itself and infer from it the validity of the moral law. Instead, in adopting a maxim we become conscious of the formal requirements of practical cognition, which are requirements of rational volition as such. I cannot become conscious of these requirements without exercising my capacity of practical volition in some way. Practical cognition is not just awareness of an existing desire together with the cognition of an instrumental principle (if you will the end, you also ought to will the necessary means) and some “built in” theoretical cognition of the required means. Practical cognition, in its full-blown sense, is acting from the cognition of a maxim as practically universal. Practical universality means that this maxim can be rationally willed by every cognizer, i.e. that the maxim as a rule of action is objectively and subjectively universal. This cognition elicits a feeling of respect for the moral law, which is the causal force of actions done “for the sake of duty”. If and only if cognition can become causally efficacious is reason practical. And we may say that, since an action done for the sake of duty is guided by an a priori cognition of the good, the action is an instantiation of a metaphysical value property. Yet the existence of this metaphysical value property remains dependent on the practical cognition of rational cognizers.

With this conception of practical cognition in place, we can come back to the Euthyphro dilemma and the question that Robert Stern says constructivists cannot answer: Why should moral realism (or an “independent order of values”), on the Kantian picture, be incompatible with autonomy (22)? If moral realism and autonomy were compatible, so the moral realist argues, moral realists could account for both objectivity and motivation. On the account I am suggesting here, the answer to Stern’s question must be that realists don’t conceive of this independent order of values as the self-consciousness of pure practical reason. And since the cognition of independent values is cognition of something given to us from elsewhere, it must be some sort of theoretical cognition, which relies on sensibility (intellectual intuition is not an option). For Kant, however, the moral law is a “fact of pure reason”, not of intuition. Therefore, moral cognition needs to be a kind of self-cognition of practical reason (the will). In willing, we become conscious of the moral law. And since what we become conscious of is our own will and not something given from elsewhere, we can say that we are autonomous in pure practical cognition. The reality of the moral law is independent of our individual choices, but it is not independent of the act of volition of rational cognizers.

My main claim against classical realism is therefore that there are no moral values without self-conscious cognizers. But this does not mean that we have to give in to the anti-realist or Rawlsian idea of weakening our conception of objectivity, or even to the tendency to embrace subjectivism. On the picture I propose, objectivity is preserved because good and evil are objects of practical cognition, which in turn is the self-cognition of practical reason. Kant is therefore a cognitivist, but he is not a moral realist, for the existence of moral facts is not mind-independent but rather depends on the self-consciousness of imperfect rational cognizers. Kant is therefore neither a moral realist in the classical sense nor an anti-realist. This fundamental contemporary distinction fails to capture Kant’s moral ontology. Kant is an anti-realist as well as an anti-anti-realist. His ontology preserves a very strong sense of objectivity without thereby conceding that moral facts exist independently of us. On his view, the existence of moral facts is brought about by the self-consciousness of pure practical reason in imperfect rational beings. If idealism is the view that the existence of objects depends on the cognition of those objects, I would like to suggest that Kant’s alternative view, which fundamentally departs from both options in contemporary metaethics, is most accurately represented as “moral idealism”.

To the extent that this is correct, the moral realist reading of Kant is inadequate: It is not the case that we somehow perceive the absolute value of human beings as ends in themselves and infer from this absolute value the validity of the categorical imperative. On the contrary, we have absolute value because we have the capacity to act from pure practical reason. This is just another way of saying that we can act from the representation of the categorical imperative.

One might object that my argument proves too little and is therefore unable to solve the Euthyphro dilemma. It ultimately runs into problems that are similar to those encountered by the moral realist, for it seems only to explain the objectivity of our moral judgments. It does not address why we ought to care more about the consistency of our own ends with those of other rational beings than about the consistency of our own ends independently of other people’s ends. Korsgaard herself raises this demand for an additional argument for why we value rationality (Korsgaard et al. 1996; Stern 2012; Schönecker 1999). But the problem is not that Kant fails to give us an argument for the connection between rationality and value. The problem is that the relation between the two has been misconceived. The demand assumes that irrationality is by itself devoid of value and that we require an explanation for why we must care about consistency. However, the categorical imperative is a principle not of consistency as such but of consistency in willing. So if we “detect” an inconsistency in willing, we think that the principle is practically impossible, i.e. impermissible, or evil. We ourselves think that it cannot be willed. We ourselves think of the principle as either only good from our private standpoint or as good unconditionally, i.e. good from the standpoint of every rational being. On the Kantian picture, we are not a multitude of isolated subjects who need to somehow come to an agreement about our particular ends. As rational beings, we all share the same formal end. Again, in pure practical reason, we are one. To ask why I should prefer what I judge to be unconditionally good to what I judge to be only of private validity is to misunderstand what it means to make these judgments. So we don’t have to be convinced by an additional argument establishing why we should be motivated not to act from principles that we cognize as practically inconsistent. The fundamental assumption Kant is making is that the idea of practical universality, the idea that all moral subjects can live in agreement, does not leave us cold.

Even if moral realists grant us this claim, they might still wonder why the capacity to act from unconditional moral laws gives unconditional value not only to our ends, but also to us. After all, our ends are objects of our faculty of volition; as such, we take them to be good in a broad sense. If it can be desired unconditionally, the end is unconditionally or absolutely (independently of my private desires) good (in a narrow sense). So why are we entitled to move from here to the additional claim that we have (or our humanity has) unconditional value? The answer is that only beings who have the capacity to make the execution of their inclinations dependent on the practical universality of their principles deserve to be treated as ends in themselves, for only their agency can be subsumed under the moral law. It would be illegitimate to judge other beings’ conduct by this law, i.e. to treat them as if they could determine themselves from pure practical reason (to treat them as ends in themselves, or as if they had unconditional value). It is not—as Korsgaard believes—as though we must value our humanity in order to value anything at all; rather, in being able to value things unconditionally (to will from the representation of laws), we deserve to be treated according to such laws, i.e. we have absolute value. To be deserving of being treated as an end in itself, one must be able not only to do something for its own sake, but also to do it under the condition that one’s end is compatible with everyone else’s ends. This is what it means for a good to be good absolutely.

Does this view amount to the absurd claim “I do value; therefore I have value”, as Rae Langton suggests (Langton 2007, 169)? It does not, for on Kant’s view a rather different claim follows: “I have the capacity to value unconditionally; therefore I have unconditional value”. Yet even this modified claim seems problematic. What is it about our capacity to value unconditionally that gives us unconditional value? According to Langton, “[w]e have no more antecedent reason to expect the creators of goodness to be good than to expect painters of the blue to be blue” (Langton 2007, 175). Langton is right about the painters, but she is wrong about the creators of goodness. Painting a wall is a case of production. In producing something, the act of production and the produced object are to be distinguished, i.e. the painter and the painted wall. The case of “creating” (absolute) goodness is more complicated. To begin with, there is an ambiguity here between creating a particular good action and creating the standard of goodness. Let us first look at a case where we bring about a particular action that is (absolutely) good. Here we are acting from duty, i.e. we are determined by the cognition of the moral law. What makes the act good is that our cast of mind is good. In this case, it is clearly right to say that the creator of goodness is also good. In adopting the good maxim, i.e. in making a valid practical judgment, I become morally good. And one might go even further; since my cast of mind is causally efficacious, one might want to add that one can even ascribe a metaphysical value property to the action as an instance of my general good volition (Bojanowski 2012; Engstrom 2013). However, this does not solve the Euthyphro problem. For even if it is true that in our actions we bring metaphysical value properties into existence, we still need to know what counts as good first. Or, to put the same point in more Kantian terms, we need to know what the proper determining ground is before we can bring the object of this determining ground into existence. The source or origin of this knowledge is what the Euthyphro problem is getting at.

So Langton might respond that her claim is directed not at particular actions but at the constructivist claim that the creation of goodness in general makes us good. Yet even in the case of the general good, Langton’s analogy does not hold. “Creation” with respect to goodness in general cannot be taken to mean that we bring a physical object into existence. And if we don’t want to fall back into voluntarism, the created good cannot be a mental object of our voluntary imagination either. In Kant, the general goodness is the same goodness every rational cognizer with a faculty of rational volition “creates”. It is created in the sense that it is not given to us from elsewhere but brought about through self-conscious activity; the good does not exist independently of this activity. And, as I’ve argued above, only beings with the capacity to take things to be good absolutely are themselves absolutely good. The painter analogy overlooks the sharp contrast between practical cognition and physical production.

4 Conclusion

Are our actions morally good because we approve of them, or are they good independently of our approval? Are we projecting moral values onto the world, or do we detect values that are already there? For Kant, the Euthyphro dilemma states a false alternative. Kant holds that our actions are good because they can be rationally willed by all rational cognizers. This solution does not run into subjectivism, because the moral law is the form of practical cognition. It does not run into the problem of moral motivation, because moral goodness is not a value property that exists mind-independently but is constitutive of the self-consciousness of practical cognizers. Moral values do not exist independently of our approval, if approval is taken to mean “rationally willed”. Moral values are not merely projected onto the world, because the moral world only comes into existence through rational volition; there are no objective moral values in the absence of practical cognizers.