When are human beings morally reponsible, and when not

Kant’s Empirical Markers for Moral Responsibility

In the Critique of Practical Reason, Kant writes that “the moral law commands compliance from everyone” (5:36, cf. 29:603).[1] In the Anthropology, he reiterates that “the law of duty . . . is present in everyone” (7:214). But in his lectures Kant also claims that “in some cases . . . [a human being] has no power of free choice, e.g., in the most tender childhood, or when he is insane, and in deep sadness, which is however a kind of insanity” (28:255). Given that for Kant free choice is necessary for moral responsibility, this implies that children and the insane are not morally responsible. In his Anthropology, Kant even allows that when “someone has intentionally caused harm,” the question can still arise “whether he is guilty of it and to what extent, so that the first thing to be determined is whether or not he was mad at the time” (7:213).[2] And in another set of lectures, Kant insists that when someone “pushes another into the water . . . and that person drowns,” there is still a question about whether such a person is morally responsible for this deed. The “push” might, for instance, have been simply the consequence of “dizziness” or some other “cause [that] was merely physical and a matter of natural necessity” (27:559).

The claim that almost all human beings are morally responsible but that some human beings (such as children) or human beings at certain times (such as when mad or dizzy) are not morally responsible seems fairly sensible. But the attempt to carve out ground for this position raises an important problem for Kant. Kant’s first Critique argues that although every event in nature is causally determined, it is nonetheless possible that the ultimate grounds of at least some events lie in free agents. In the first Critique, Kant’s defense of freedom is extremely limited. As he says, “we have not been trying to establish the reality of freedom, . . . [nor even] the possibility of freedom . . ., [but only] that nature at least does not conflict with causality through freedom” (A558/B586). In the Groundwork and the second Critique, Kant goes further, seeking to show the reality of freedom, at least in the case of human agents. The argument of the Groundwork argues from the consciousness of the idea of freedom to participation in an intelligible realm and thus to actual freedom (4:452).[3] By the time of the second Critique, Kant seems to have rejected this argument in favor of a more straightforward regressive proof of freedom as the condition of the possibility of moral responsibility. As Kant puts it there, one “judges that he can do something because he is aware that he ought to do it, and cognizes freedom within him, which, without the moral law, would have remained unknown to him” (5:30).

Kant’s shift to this regressive argument in the second Critique has been criticized for not adequately dealing with the skeptical concerns that prompted his earlier attempts to find a non-moral argument for freedom.[4] But the shift in argumentative strategy also raises a specific problem for identifying moral responsibility. Kant seems to assume in the second Critique that the ascription of moral responsibility to an entity is trivial, at least absent any skeptical doubts. But often it is not. As Kant points out, in the cases of children, the insane, and even those in “deep sadness,” it becomes unclear where to draw the line. One may extend these concerns about moral responsibility to other human cases, and even to animals. On what grounds, for instance, do we justify holding most human beings morally responsible and not chimpanzees or dolphins?[5] The argument of the first Critique, showing that natural necessity does not conflict with freedom, works just as well for these animals as for humans. One cannot use Kant’s strategy in his early ethics lectures of distinguishing cases of responsibility from those in which the cause is “a matter of natural necessity” (27:559), because according to Kant’s transcendental idealism every human deed is a matter of natural necessity. And one cannot – at least by the time Kant rejects the arguments of the Groundwork – argue from a person’s (or animal’s) transcendental freedom to their moral responsibility because there is no way to prove that any being is transcendentally free except from the conditions of possibility of moral responsibility. So how can Kant distinguish between those who are morally responsible and those who are not? And even in cases of moral agents, how can one distinguish acts or dispositions for which one is morally responsible from the sadness and madness that absolves one of guilt?

In answering these questions, it is important to avoid the temptation of the Groundwork. The Kantian should not seek to reason from a particular empirical psychology to a non-moral proof of freedom as a ground of moral responsibility. Kant does acknowledge that the moral law, and with it moral responsibility, “would be analytic if the [transcendental] freedom of the will were presupposed” (5: 31, cf. 4: 447), but this sort of freedom cannot in principle be experienced by human beings. By the second Critique, Kant gives up on any attempt to provide a proof of freedom independent of human moral responsibility, arguing instead that one is immediately conscious of one’s responsibility to obey the moral law.

Consciousness of this fundamental law may be called a fact of reason because one cannot reason it out from antecedent data of reason, for example, from consciousness of freedom . . . and because it instead forces itself upon us as a synthetic a priori proposition that is not based on any intuition. (5:31).

Given this “fact of reason,” one can show that human beings are transcendentally free. But there is no proof of this fact itself. Even in the first Critique, Kant recognizes that although “the question of the possibility of freedom does indeed assail psychology” yet “since it rests merely on dialectical arguments of pure reason, its solution must be solely the business of transcendental philosophy” (A535/B563). And Kant is even clearer in a metaphysics lecture from the early 1790’s,

Freedom cannot be proven psychologically, but rather morally. Through morality I consider a human being not as a natural being, as object of the senses, but rather as intelligence, as object of reason. If I wanted to prove freedom psychologically, then I would have to consider a human being according to his nature, i.e., as a natural being, and as such he is not free. (28:773, cf. 28:682)

Because even inner experience is necessarily structured by the category of causation, there is no empirical psychology that can justify moral responsibility on its own. Moreover, given Kant’s transcendental idealism, there is no empirical psychology that is incompatible with freedom. Paul Guyer has even suggested, on these grounds, that “the subjective state of one’s feelings” can, perhaps even directly, “reflect the moral choices of one’s will” (Guyer 1993: 367).

Nonetheless, although Kant’s transcendental idealism does not force him to adopt any particular empirical psychology, he in fact does develop a very specific account of human action at the empirical level, and this empirical picture is constructed in a way that highlights certain empirical features of human action that correlate with human freedom. In ethics lectures as late as winter of 1793, for example, he describes a “visible spontaneity” in certain actions that is “an essential criterion of freedom” (27: 505). Kant quickly makes clear here that this visible spontaneity is not the transcendental freedom that is a condition of the possibility of freedom; immediately after referring to “visible spontaneity” in human nature, Kant raises the possibility that actions proceeding from this spontaneity might be “grounded, simultaneously, in the time preceding” such that “unconditioned self-activity would not be present in it” (27:505). Because “it was this [unconditioned self-activity] that was demanded of man qua noumenon or intelligible being, . . . only as an intelligible being does he emerge completely from the world of the senses . . . . Freedom, therefore, cannot be made comprehensible” (27:505). Still, Kant insists that this visible spontaneity is an important “criterion” [Criterium] of freedom. And in his anthropology Kant refers to character, an empirically recognizable capacity of a human being that is associated with one’s visible spontaneity, as “a mark [Merkmal] of a rational being” (25: 1156) or even a “distinguishing sign [Unterscheidungseichen] of a rational being endowed with freedom” (7:285).[6] Visible spontaneity is neither a necessary nor a sufficient condition for transcendental freedom, but it can still be an important empirical criterion, sign, or marker of it.

In his Groundwork, and in greater detail in his second Critique, Kant gives even more detailed accounts of empirical correlates to moral freedom, describing what the moral law empirically “effects in the mind insofar as it is an incentive” (5:72, cf. 4:400). In his Metaphysics of Morals, when Kant explains “the relation of the human mind to moral laws” (6:211), he even draws on his empirical account of human psychology to explain where to situate moral motives in human psychology. And in his Anthropology, when he raises the possibility that a person might not be morally responsible for her actions, Kant insists that this question is “purely psychological,” to be solved by determining “whether the accused was in possession of his natural powers of understanding and judgment” (7:213-4, my emphasis). Thus although Kant insists that “by empirical psychology . . . we should know ourselves merely in the world of sense . . . . [and] therefore morality is the sole means of obtaining consciousness of our freedom” (27:506), he nonetheless holds that empirical psychology provides empirical markers that can be used to distinguish cases in which freedom is present from those in which it is not.

The first task of the rest of this paper will be to briefly explain Kant’s empirical psychology insofar as this bears on moral responsibility. I then argue that Kant’s empirical markers for moral responsibility are not sufficient conditions for moral responsibility, primarily on the grounds that they are empirical characteristics that are subject to natural laws. This causal determination, combined with Kant’s insistence in the second Critique that transcendental freedom is a condition of the possibility of moral responsibility, shows that these criteria do not constitute a proof of moral responsibility. In this context, I describe two situations within which one could meet all of Kant’s empirical criteria and still fail to be morally responsible. Next, I argue that these empirical markers are not necessary conditions of moral responsibility. Given Kant’s transcendental idealism, even an entity that lacks these empirical markers could be free and thus morally responsible, although as a matter of fact Kant thinks that none are. Finally, I use Kant’s account of the markers of moral responsibility to explain why Kant thinks that people cannot generally be held morally responsible for their emotions (see 4:339). Drawing on my claim that empirical markers are not necessary conditions, however, I discuss the possibility of a revised Kantian moral theory within which one could be held morally responsible for emotions, and I raise an issue for further work within neokantian ethics.

Before turning to Kant’s discussions of empirical psychology, it is important to note two difficulties with presenting any clear account of the empirical markers of moral responsibility. First, throughout Kant’s discussion he uses the same terms to refer to both noumenal bases and phenomenal causes of human action. Of these, the noumenal bases are in fact necessary (and in some cases sufficient) conditions of moral responsibility, while the phenomenal causes are merely markers for that responsibility. For example, within Kant’s practical philosophy, he discusses a free noumenal power of choice (Willkühr) combined with a pure practical reason (Wille) that legislates for that power of choice. And Kant makes clear throughout his moral philosophy that this (transcendental) freedom is, as he puts it in the second Critique, the ratio essendi of the moral law. In that sense, a free Willkühr is a necessary and perhaps a sufficient[7] condition of moral responsibility. But Kant also discusses the free Willkühr as simply a capacity of certain organisms – human beings – to have desires that are caused by certain sorts of cognitions. And in this context, neither “freedom” nor the “Willkühr” are necessary or sufficient for moral responsibility; they are simply (as we will see) empirical markers.[8]

What makes this terminological ambiguity even more confusing is that early in his thinking Kant seems to have thought that the freedom necessary for moral responsibility could be established within empirical psychology. In a lecture from the 1770s, for example, Kant claims that “practical or psychological freedom . . . is treated of in empirical psychology, and this concept was also sufficient enough for morality” (28:267). Thus at least in this early lecture, the freedom that Kant discusses within his empirical psychology is both an empirical property of human beings and sufficient for – rather than merely a marker of –moral responsibility. And although Kant gives up this argumentative strategy in later lectures, the terminological confusion remains. What is more, as Kant develops his transcendental idealism, he often uses the discussion of (empirical) freedom in his empirical psychology as a starting point for discussing his transcendental philosophy. The result of these shifting views and ambiguous terminology is that it is often difficult to distinguish the perspective from which Kant is speaking at any given time, and this makes it look as though what are really only markers of moral responsibility are necessary or sufficient criteria of it. Still, it is possible to distinguish between Kant’s empirical accounts and his transcendental ones at least to the extent that a reasonable Kantian view can be reconstructed.

a) Empirical psychology and moral responsibility

Kant’s empirical psychology is organized around three basic faculties of the soul – cognition, feeling, and desire. Of these, the faculty of desire is the most important for understanding the empirical markers of moral responsibility because “all desires have a relation to activity and are the causality thereof” (25:1514). Within each of his three faculties, Kant distinguishes between several basic powers, grouping these into “higher” and “lower” faculties of cognition, feeling, and desire. With respect to cognition, the higher faculty includes reason, the understanding, and judgment; the lower faculty includes the senses and imagination. With respect to the faculty of desire, Kant distinguishes the higher and lower faculties on the basis of the faculty of cognition that causes the relevant desire. For Kant, every desire is caused by some cognition, but one can distinguish between desires with causes that “lie . . . in the understanding” and those with causes that lie “in the sensibility” (29:1014). The former are “motives” and belong to the higher faculty of desire; the latter are “stimuli” and belong to the lower faculty (29:1015).

For Kant, the “visible spontaneity” that marks moral responsibility is associated with the higher faculty of desire. As Kant explains, “the concept of freedom rests on this: namely the faculty of a human being for determining oneself to action through motives” (29:1016). And because the distinguishing feature of the higher faculty of desire is its determination by the higher faculty of cognition, the higher faculty of cognition becomes an important distinguishing mark of those endowed with moral responsibility. Thus Kant says, “Reason is the persisting condition of all voluntary actions under which the human being appears” (A553/B581), and Kant uses the “understanding” to distinguish the “power of free choice <arbitrium liberum>,” which can occur only with human beings, from the arbitrium brutum of animals (28:588).[9] Likewise in the Anthropology, Kant insists that courts must refer the question of whether a criminal should be held morally responsible to empirical psychologists[10] because this issue rests on “the question of whether the accused at the time of his act was in full possession of his natural faculty of understanding and judgment” (7:213).

This “natural faculty of understanding” does not involve anything specifically moral and need not even be purely rational. Although all “higher” desires have “grounds of determination . . . [that] lie . . . in the understanding” (29:1014), these desires can be “either pure or affected” (29:1015). As Kant explains,

The intellectual impelling cause is either purely intellectual without qualification <simpliciter talis, mere intellectualis>, or in some respect <secundum quid>. When the impelling cause is represented by the pure understanding, it is purely intellectual, but if it rests on sensibility, and if merely the means for arriving at the end are presented by the understanding, then it is said to be in some respect <secundum quid>. (28:589)

When a desire is impure but still associated with the higher faculty of desire, one acts on the basis of a principle of the understanding that is directed towards fulfilling some lower desire. But although such impure higher desires amount to merely hypothetical imperatives, they still relate to the higher faculty of desire. In that sense, even a capacity to act on hypothetical imperatives is a marker of moral responsibility.[11]

In the Anthropology, Kant seems to associate moral responsibility merely with the capacity for action from a higher faculty of desire, and he suggests that for practical purposes, this issue reduces to the question of whether someone has a “natural faculty of understanding.” But sometimes Kant suggests that the capacity for one’s faculty of desire to be determined by pure reason is also an important marker of moral responsibility. That is, one who is morally responsible will have at least the capacity for impelling causes that are purely intellectual.[12] In the Metaphysics of Morals, for example, Kant shifts from his standard definition of the free Willkühr, or arbitrium liberum as the faculty of desire that is determined by the understanding, broadly construed. There he claims that “That choice which can be determined by pure reason [Kant’s emphasis] is called free Willkühr” (6:213). By contrast, in a lecture probably given in 1790-1, Kant explains that “power of free Willkühr <arbitrium liberum> can occur only with human beings, who have understanding” where this understanding can be “either pure or affected” (28:588, but cf. 28:677). The shift to pure reason in the Metaphysics of Morals might reflect Kant’s concern with defining the “faculties of the human mind” insofar as these relate “to moral laws” (6:211). And this suggests that a capacity for action on the basis of pure reason might be an important criterion of moral responsibility in human agents.[13]

This capacity to be motivated by the moral law itself also seems to play a role in Kant’s accounts of moral psychology in the Groundwork and the second Critique. When Kant discusses “respect for the moral law” in these contexts (cf. 4:400, 5:72-89), he seems to lay out an empirical cause of morally good action. In the second Critique, for example, he explains that his account will show “what happens to the human faculty of desire as an effect of [the moral law as a] determining ground” of one’s actions (5:72). The elaborate account offered there of the moral law as an incentive is congruent with Kant’s more general accounts of the pure intellectual faculty of desire, whereby pure practical reason can cause various actions. In both the Groundwork and the second Critique, however, Kant’s discussion of respect is not primarily an account of the empirical correlates of moral responsibility. Rather, it is an account of the experience of morally good action. Thus although these texts provide important empirical discussions of the experience of acting from duty, they do not explicitly address the criteria for determining whether an entity is morally responsible. Nonetheless, it is natural to read them as offering a supplement to Kant’s general focus on the higher faculty of desire, and one might reasonably think that if a person lacks the capacity to feel this kind of respect, one cannot rightly hold them responsible.

We have, then, the beginning of an answer to the question of when a human being (or other animal) is morally responsible. Insofar as the cause of an action can be traced via one’s desires to cognitions that lie in the understanding (in the broad sense), one might be morally responsible. To this Kant sometimes adds a capacity for one’s pure reason to be a cause of desires that give rise to actions, a capacity connected with one’s consciousness of the moral law. Either empirical account explains why children and the insane are not morally responsible. Children – at least very young children – do not have sufficiently developed rational capacities to act on the basis of principles of the understanding or reason. And at least for certain cases of insanity, the possibility for one’s desires to be shaped by one’s understanding has been lost.

This account also can be extended to cover particular deeds[14] of otherwise morally responsible agents. One who drowns another due to dizziness is not morally responsible even if the person has an otherwise properly functioning faculty of understanding because the dizziness itself is something over which the understanding does not and could not have any causal influence, and dizziness has the effect of causing action without allowing for influence by the understanding. It disrupts the pathway from higher cognition to higher desire. Kant even classifies a whole sphere of mental weaknesses – affects – that may cause one to act or fail to act without moral responsibility by causing one to act without the influence of the higher cognitive faculties.[15] Kant explains, for example, the way that shock can incapacitate without any moral responsibility attaching to one’s failure to act:

One sees a child fall into the water, who one could save, however, through a small aid, but one is so shocked that one thereby cannot do anything. Shock anaesthetizes someone such that one is thereby totally incapable of doing anything. (25:592)

In this case, one does not have more responsibility because the proper functioning of one’s higher cognitive faculties, and thereby of one’s higher faculty of desire, has been impeded. In general, then, Kant can answer the question of what warrants ascribing moral responsibility in particular cases by appealing to the empirical fact that the relevant agents have properly functioning higher cognitive faculties.

One might add to this a requirement that one have the capacity for actions to be motivated by the pure higher cognitive faculties, but generally for Kant the work of distinguishing cases rests merely on the presence of a higher faculty of desire, and that in turn on the presence of a functioning understanding. The capacity for action to be caused by pure reason does not play a prominent role in the practical task of explaining who can be excused from moral responsibility. Thus the ambiguity in Kant’s own discussions of determination by pure reason is not particularly important in the concrete application of empirical tests for moral responsibility.[16]

b) Empirical markers are not sufficient for moral responsibility

In the last section, I argued based on Kant’s empirical psychology that the presence of higher cognitive faculties is an important criterion for moral responsibility. My justification of this empirical marker for moral responsibility, however, might seem to have proven too much. On the account that I have offered, an agent is morally responsible if they are motivated by various higher cognitions and can be motivated by the moral law. I have insisted that this “criterion” of agency is merely a marker for moral responsibility; it is neither a necessary nor sufficient condition. But one might question whether an agent could ever be morally responsible if they cannot be motivated by the moral law, or if their actions are truly instinctual, caused by immediate sensations with no endorsement by higher faculties of cognition and desire. Similarly one might think that someone who meets these criteria must be responsible. That is, one might think that these “markers” must be necessary and sufficient conditions of moral responsibility. In this section, I take up the question of whether these markers are necessary for moral responsibility. In the next, I address the issue of sufficiency.

In his lectures on ethics, Kant is explicit that higher faculties are not sufficient for moral responsibility. There Kant says,

The ground of the fact that man is an accountable being, lies

a. not simply and solely in the fact that he is a rational being; accountability will, indeed, be founded a posteriori on that, but a priori is can still be separated therefrom. The idea is acceptable a priori that man, by virtue of his rational capacity, can reflect upon the grounds and consequences of his action, without his morality having to be connected with that….

b. absolutely necessary in addition, that he act with freedom, indeed it is only when considered as a free being that he can be accountable. (27:559)

As Kant makes clear here, when it comes to a posteriori ascriptions of moral responsibility, one can turn to the fact that human beings are rational. This reflects the fact that rationality is a legitimate empirical marker of moral responsibility. But the empirical fact that one is rational is insufficient to justify moral responsibility philosophically, because one could be rational without being accountable, if one lacks freedom (which here refers to transcendental freedom).[17] And precisely because these empirical markers are empirical, they cannot provide any direct evidence that a person has that transcendental freedom that is the necessary condition of the possibility of moral responsibility.

The basic argument against considering these empirical markers to be sufficient conditions of moral responsibility has three steps. The first step is to show that the markers of moral responsibility that I have discussed are empirical characteristics that fit into a series of natural causes and effects.[18] Kant makes the empirical nature of these markers clear throughout his writings. In the Anthropology, the “understanding” that indicates culpability is a “natural faculty” (7:213). And in the first Critique, Kant is quite explicit about the general point that the “causality of reason in the determination of the will” that he associates with empirical freedom is “one of the natural causes” (A803/B831). He adds that “even though it is reason, it must nevertheless exhibit an empirical character” (A549/B577) and raises the possibility that “reason is itself determined by further influences” (A803/B831). And in a 1793 lecture on ethics, Kant insists,

Even one’s reason, as subjected to the laws of nature, can be considered devoid of all freedom . . . . Man is not set free from the mechanism of nature by the fact that in his action he employs an actus of reason. Every act of thought or reflection is itself an occurrence in nature . . . . So the fact that a man is determined to action on grounds of reason and understanding does not yet release him from all mechanism of nature. (27:502-4)

Although higher faculties of cognition and desire are important empirical indicators of moral responsibility and even constitute a kind of empirical freedom, they are still part of a series of causes and effects in nature.

The empirical nature of the higher faculty of desire is confirmed by Kant’s description of various empirical influences on one’s choices. Throughout his anthropological writings, Kant points out ways that one can influence the decisions of others, including influences on their higher faculties of desire.[19] In addition to general suggestions, Kant is particularly concerned with how to influence others for their moral betterment, claiming that knowledge of human nature “is . . . indispensable and manages great uses . . . with respect to the influences on morals and religion, that through this knowledge one can give these duties the power of inclinations” (25: 1437). And in his lectures on ethics, Kant even explains how “a person may be compelled to duty by others” (27:521).[20] Insofar as they are part of Kant’s empirical account of human action, even actions that proceed from the higher faculty of desire can be explained by reference to empirical causes.[21]

And Kant gets quite specific about various empirical influences on human behavior, including influences on the development of particular patterns of intellectual desire. For example, politeness in social interactions promotes “loving the good” (25:931, cf. 6:473, 7:151-3)[22] and “the beautiful prepares us to love . . . without interest” (5:267).[23] “Unsocial sociability” is a “means that nature employs to bring about the development of all our predispositions [Anlagen]” (8:20).[24] In particular, this unsocial sociability can “transform the primitive natural predisposition [Naturanlage] for ethical discrimination into definite practical principles” (8:21). Even political stability provides a context within which “the citizens’ inclination to violence against one another is powerfully counteracted by a greater force, namely that of government, . . . [by which] the development of the moral predisposition to immediate respect for right is actually greatly facilitated” (8: 375n).[25] In all of these cases, social and cultural influences give rise to various patterns of intellectual desire, and these patterns can in turn explain individual human choices.[26] For Kant, the higher faculties of cognition and desire are not free from determination by empirical influences.

Given that the empirical markers of moral responsibility – the faculties of understanding and the higher faculty of desire – can be explained naturalistically, there are two further steps to show that these markers cannot be sufficient conditions for moral responsibility. The next step is to argue that anything that fits into a natural series of causes and effects cannot be equivalent to transcendental freedom. This is quite straightforward, because Kant defines transcendental freedom as “independence from everything empirical and so from nature generally” (5:97). Finally, one must argue that transcendental freedom is a necessary condition of moral responsibility. Although this is complicated by the fact (noted above) that Kant changes his mind about the conditions of possibility of moral responsibility, by the time of the second Critique Kant makes quite clear that only transcendental freedom is sufficient for moral responsibility.[27] As he says there, “without [transcendental] freedom . . . no moral law is possible and no imputation in accordance with it” (5:97). Because the empirical markers of moral responsibility do not establish this transcendental freedom, they are insufficient proof of moral responsibility, even if, as Kant says, they are sufficient a posteriori criteria.

Even with this abstract account of why empirical markers are not sufficient for moral responsibility, it might just seem implausible that one who is empirically capable of being motivated by the moral law is not morally responsible. Fortunately, Kant considers at least two hypothetical cases within which human beings would not be morally responsible despite a psychological account of action that includes higher faculties of cognition and desire. First, in the first Critique, Kant considers the possibility that transcendental idealism is false, that there is an incompatibility between nature and freedom. He first reiterates that “that morality necessarily presupposes freedom” but then raises the possibility – contrary to fact – “that speculative reason had proved that freedom cannot be thought at all” (Bxxix). In other words, speculative reason might have shown, and many people both in Kant’s day and our own think it has shown, that given natural necessity, freedom is simply unthinkable. Kant develops his transcendental philosophy in part to show that this inference from causal necessity to the impossibility of freedom is unfounded, but he here entertains the possibility that his arguments for transcendental idealism fail. And in such a case, he claims, “freedom and with it morality . . . would then have to give way to the mechanism of nature” (Bxxix). As he says later in the first Critique, “if appearances are things in themselves, then freedom [and by implication morality] cannot be saved” (A536/B564). Thus even if human beings have the psychological capacity to be empirically determined by higher faculties of cognition and desire, and even if human beings can sometimes be empirically determined by pure rational cognitions, unless this empirical determination is itself grounded in transcendental freedom, human beings are not morally responsible.

In the Religion, Kant raises a second scenario within which human actions might empirically act from higher faculties but within which humans would not be morally responsible. There Kant addresses the problem that human beings are “radically evil,” corrupt at the level of our most fundamental maxims. He suggests that “some supernatural cooperation is . . . needed for his becoming good or better” (6:44), but in order to save moral responsibility, Kant must carefully restrict the scope of this intervention. He argues,

The concept of a supernatural intervention into our moral though deficient faculty . . . – this is a transcendent concept, merely an idea of whose reality no experience can assure us. – But even to accept it as idea for purely practical intent is very risky and hard to reconcile with reason; for what is to be accredited to us as morally good conduct must take place not through foreign influence but only through the use of our own powers. (6:191)

Here Kant entertains the possibility that God might bring about a moral shift in one’s fundamental maxims through an act of grace, but he raises the practical problem that insofar as God – a “foreign influence” – causes this shift, it cannot be accounted to oneself and thus cannot constitute true moral goodness. As a result, Kant insists that “the human being must . . . make himself antecedently worthy of receiving” grace (6:44). For the purposes of this paper, I am not interested in the specific dynamics of Kant’s account of grace. What is important here is only that if God caused a change in one’s fundamental moral maxims, the effect of this change would be that one would more consistently act from principles of pure reason. One would satisfy the empirical criteria Kant sets out for moral responsibility. But because the ultimate ground of one’s actions would lie in God, rather than in one’s own transcendental freedom, one would not in fact be morally responsible, in that one’s morally good conduct could not be accredited to one.[28]

Both of these accounts have in common that Kant’s empirical account of human agency is left unchanged, but that account is shown to be ultimately grounded not in human freedom but in something else – natural laws or God’s grace.[29] In these cases, despite the fact that human beings can act on the basis of rational principles, and even on the basis of pure principles of practical reason, such action is not free because it is not noumenally free. And this lack of transcendental freedom is sufficient for Kant to deny moral responsibility without any change in his empirical psychology. Thus the empirical psychology that serves as a marker for moral responsibility is not sufficient for moral responsibility.[30]

c) Empirical markers are not necessary conditions of moral responsibility

Even if no empirical features of human psychology are sufficient for showing that one is morally responsible, one might think that at least some empirical features are necessary for moral responsibility. That is, one who acts on the basis of principles and has the capacity to act on pure principles of reason might not be morally responsible if their behavior is ultimately determined by God or natural laws, but there is no way to hold morally responsible a person who cannot act on the basis of principles at all.

As far as I know, Kant never explicitly discusses the possibility of an entity that is morally responsible but lacks an empirical psychology that includes a higher faculty of desire and the capacity for this desire to be determined by pure principles of practical reason. (Even God, it seems, would meet these two criteria.) Nonetheless, Kant’s transcendental idealism and his account of transcendental freedom open the possibility for ascribing moral responsibility to entities regardless of their empirical psychology.

Kant’s argument for transcendental freedom in human beings shifts between the Groundwork and the second Critique, so here I will focus on the structure of the argument in the second Critique. That argument takes place in the context of the transcendental idealism that Kant defends in the first Critique. In particular, the Third Antinomy of the first Critique raises the possibility that “there is no freedom” because “everything in the world happens solely in accordance with the laws of nature” (A445/B473). Kant responds to this possibility by claiming that although “every effect in the world must arise . . . from nature,” yet there is still the possibility that “both [nature and freedom], each in a different relation, might be able to take place simultaneously in one and the same occurrence” (A536/B564). Kant develops his account of this possibility in terms of the notion of “character.” As he says,

Every effective cause must have a character, i.e., a law of its causality, without which it would not be a cause at all. And then for a subject of the world of sense, we would have first an empirical character, through which its actions, as appearances, would stand through and through in connection with other appearances in accordance with constant natural laws, from which, as their conditions, they could be derived . . .. Yet second, one would also have to allow this subject an intelligible character, through which it is indeed the cause of those conditions as appearances, but which does not stand under any conditions of sensibility and is not itself appearance . . ..

In its empirical character, this subject, as appearance, would thus be subject to causal connection . . .. But in its intelligible character . . ., this subject would nevertheless have to be declared free . . .. (A539/B567)

Kant does not think that his distinction between empirical and intelligible character is sufficient to prove that the “subject” is in fact free. He does not here establish “the reality of freedom . . . [but only] that nature at least does not conflict with . . . freedom” (A558/B586). But what is crucial for my purposes is that the account that establishes that nature does not conflict with freedom does not depend on any particular account of the sorts of natural causes at play. Kant’s point is simply that natural necessity of any kind can be grounded in an intelligible character that can be transcendentally free.[31]

In that context, there is nothing preventing the ascription of an intelligible character, and with it transcendental freedom, to any empirical objects at all. To use Hume’s famous example, it is possible – in a very weak sense – that the “sapling . . . which . . . at last overtops and destroys the parent tree”[32] is transcendentally free (and thus potentially both morally responsible and guilty of parricide!). The invocation of “character” even highlights this possibility because Kant is explicit that “every effective cause must have a character” (A539/B567), including the causes governing the development of trees as much as the causes governing human behavior.[33] Of course, the arguments in the first Critique merely show the compatibility of freedom with natural laws. They establish freedom for neither saplings nor human beings. But the claim of compatibility works equally well for both cases.

Kant’s positive argument for human freedom comes in the second Critique.[34] In perhaps the most intuitive statement of the structure of his argument, Kant explains that one “judges, therefore, that he can do something because he is aware that he ought to do it” (5:30). This argument begins from the fact that one “ought to do” things. From that “fact” of moral obligation – the moral law itself – one judges that one is free because freedom is a condition of the possibility of moral responsibility.[35] Because the moral law applies to human agents, humans must be morally responsible. Therefore, humans must be transcendentally free.

Now if we assume that this argument works in the case of human beings, we are not committed to saying that it will works in the context of the parricide sapling. In that sense, Kant is not committed to the apparently absurd conclusion that saplings are free and thus blameworthy. But the reason that the argument does not apply to the sapling raises some concerns about Kant’s overall argument. Because saplings fail to exhibit any of Kant’s empirical markers for moral responsibility, saplings are not morally responsible.[36] And because the basis of ascribing transcendental freedom to human beings is that they are morally responsible, there is no basis for ascribing it to trees. But now it should be clear that this argument poses a problem for the claim that empirical markers can be shown to be necessary for moral responsibility. Kant can show that the sapling is not transcendentally free only by assuming that his empirical markers are reliable criteria for moral responsibility. If anyone challenges this assumption, none of Kant’s arguments in the first or second Critique will be sufficient to prove that one cannot hold saplings responsible, and the overall structure of those Critiques will even provide a systematic account of how to make sense of such moral responsibility.[37] Thus any argument to show that Kant’s empirical markers are necessary for moral responsibility will be circular.

d) Moral responsibility and the emotions

Of course, Hume was correct to say that saplings cannot be held morally responsible, and thus that “parricide” makes no sense to describe the sapling killing its parent tree. And very few philosophers, in Kant’s day or today, would claim that moral responsibility should be extended to trees. But philosophers in both Kant’s day (such as Adam Smith) and today (such as Paul Guyer and Martha Nussbaum) claim that human beings should be held responsible for their emotions. Paul Guyer in particular has used an argument similar to the case of the sapling to argue that even within a Kantian account, there is no reason not to hold people responsible for their emotions. As Guyer explains in Kant and the Experience of Freedom,

there is no metaphysical reason why the effect of . . . an act of [free] choice must be confined to anything like the phenomenal manifestation of reason itself – that is to say, why it must manifest itself only in the strength of reason, as contrasted with inclinations in the phenomenal sphere. (Guyer 1993: 362-3)

As a result, one can open room for a Kantian account within which “the subjective state of one’s feelings” can “reflect the moral choices of one’s will” (367). Guyer’s argument is similar to the argument I have given for the possibility of sapling moral responsibility. Guyer appeals to the fact that for Kant, “especially in the works from the Critique of Practical Reason to the Religion, this ‘intellectual ground’ [of one’s actions] is understood as a complex event consisting of a free but ultimately inscrutable act of choice,” an act that is “noumenal” and one that Kant does “not identify . . . with reason as we may understand it through our phenomenal conception of reason” (Guyer 362). Thus Guyer seems to have room here for including “the entire phenomenon of the agent’s character” (364) and perhaps even “the entire phenomenal world” (363) as an expression of this inscrutable choice.

Kant’s response to Guyer will be the same as his response to Hume. Just as Kant does not hold saplings morally responsible, he does not hold human beings morally responsible – or at least not directly responsible – for their passing feelings, and certainly not for the state of the world around them. When Kant says, “love as an inclination cannot be commanded” (4:399), he takes it out of the realm of the argument for freedom offered in the second Critique. Because love as an inclination is not something that one “ought” to feel, it is not necessarily to ascribe it or its absence to the transcendentally free self. And Kant can explain why “love as an inclination” is not within the realm of moral responsibility because inclinations, for Kant, are habits of the lower faculty of desire.[38] Love as an inclination is not sufficiently intellectual to count as a higher desire, and thus it does not fall under Kant’s empirical criteria for moral responsibility. Just as one is not responsible for sudden dizziness, one is not responsible for love. Thus although Guyer’s claim that one might be morally responsible for one’s feelings is consistent with Kant’s transcendental philosophy, it is inconsistent with Kant’s specific account of the empirical criteria for moral responsibility.

(Kant does discuss one class of emotions – the passions – that are principled and thus can be ascribed to one’s freedom. Because these passions “can be coupled with the calmest reflection” (7:265), such that “passion is always associated with purposes of reason” (7:266) one can be held morally responsible for one’s passions. Thus Kant points out that “one cannot attribute passions to mere animals” (7:266), and he refers to them as “properly evil” (6:408). In this case, Kant’s account of the empirical markers for moral responsibility gives him a way to ascribe moral responsibility for one narrow class of emotions. Unfortunately for those seeking to find a positive place for emotions in Kant’s account, all passions are evil.)

Kant’s general strategy of delimiting the scope of moral responsibility will work for any aspects of the empirical world that do not have Kant’s empirical markers for moral responsibility. Because Kant does not hold animals or very young children morally responsible, he need not ascribe freedom to them. And because Kant does not hold people morally responsible for their eye color, passing emotions, or environment, he need not ascribe those aspects of their empirical character to a transcendentally free ground. But Guyer poses a significant difficulty with Kant’s account. Presumably Guyer wants to root feelings in one’s transcendental freedom because he does think that human beings are morally responsible for their feelings. And if humans are morally responsible for them, then they can rightly be ascribed to one’s transcendental freedom. Just as in the sapling case, Kant seemingly has no non-question-begging response to Guyer. The argument of the first Critique shows that no aspect of the empirical world is out of bounds as a locus of transcendental freedom. And the argument of the second Critique shows that one can reasonably believe that agents have transcendental freedom because and to the extent that one holds them morally responsible. So if Guyer wants to hold people morally responsible morally responsible for their emotions, Kant’s systematic philosophy provides an account of how to do that. And unlike the case of Hume’s patricidal sapling, Guyer’s proposal that one hold people responsible for their emotions is a live option in contemporary philosophical ethics.

There is a reason that Kant does not have an immediate response to Guyer. In a footnote at the beginning of the Critique of Practical Reason, Kant responds to a critic who accuses him of having nothing new to say about ethics. Kant asks,

Who would even want to introduce a new principle of all morality and, as it were, first invent it? Just as if, before him, the world had been ignorant of what duty is or in thoroughgoing error about it. (5:9n)

Kant sees his moral philosophy, and arguably his transcendental philosophy as a whole, as providing a rigorous philosophical defense of moral claims that “the world” has long known.[39] This humility about his own project can be ascribed, at least in part, to Kant’s reading of Rousseau. Kant describes his encounter with Rousseau in a now famous journal entry:

There was a time when I . . . despised the rabble who knows nothing. Rousseau set me right. This blind prejudice vanishes; I learn to respect human nature. (20:44)

As I have described Kant’s argument for human freedom, it is an account that defends and justifies what Kant takes to be common human convictions about the nature of moral responsibility and obligation. And Kant’s account of empirical markers for moral responsibility is, similarly, an attempt to provide a systematic philosophical account that captures what “the rabble” already thinks about moral responsibility. Thus Kant would defend his claim that one cannot be held responsible for (most) emotions by appealing to common sense. And the problem that Guyer raises is that common sense may not be as common as Kant supposes. In cases where there is no clear common sense view about moral responsibility, Kant’s criteria for moral responsibility are just one proposal among many. His account fits in a particularly natural way with the transcendental idealism he develops to deal with certain challenges to moral responsibility. And it fits well with a wide variety of common sense views about the nature and scope of moral responsibility. But other proposals, such as Guyer’s, can also be made to fit Kant’s overall philosophical structure, and these alternatives may be equally capable of reflecting widespread (and perhaps changing) attitudes towards responsibility.

e) Conclusion

The first purpose of this paper has been to explain Kant’s empirical criteria for moral responsibility. The most important criterion is a higher faculty of desire, which is dependent upon a higher faculty of cognition. That is, Kant claims that those with a properly functioning understanding are morally responsible. Sometimes he suggests that a capacity for one’s faculty of desire to be determined by pure intellectual cognitions (the moral law itself) is important for moral responsibility, although Kant seems ambivalent about the importance of this criterion, and it seems relatively unimportant in practice. In addition to pointing out these criteria, I have emphasized that they are merely signs or markers of moral responsibility, rather than necessary and/or sufficient conditions. In the final section, I have raised a potential problem for Kant arising from the modesty of his claims about these empirical criteria. In cases of disagreement, Kant can offer an account of how to distinguish whether someone is responsible, but he cannot defend that account except by appeal to ordinary intuitions about moral responsibility.

This problem is not devastating for Kant’s philosophy, nor even for his account of empirical markers for moral responsibility. Arguably, the compatibility of Kant’s overall idealism with different accounts of the empirical criteria of moral responsibility is a strength of that account. Kant’s philosophy does not rise or fall with any particular empirical psychology. Moreover, the flexibility of Kant’s overall transcendental philosophy provides important opportunities for contemporary neokantians to develop authentically Kantian theories of moral responsibility (and even of transcendental freedom) that are not identical to Kant’s account. Paul Guyer provides an excellent beginning to such a theory, explaining how one could, consistent with Kant’s overall philosophy, defend the claim that people are directly responsible for their emotions. One might develop similar theories to explain how one can be morally responsible for the way one perceives the world or to explain how certain kinds of animals can be held morally responsible.

But with this opportunity comes a special challenge. Kant’s philosophy needs to be supplemented with an account of how one can arbitrate between competing common sense views of moral responsibility. Given several different possible Kantian theories of the empirical criteria of moral responsibility, one needs a way to distinguish between them. And Kant’s idealism has cut off what might seem the most obvious way. One might claim that any action that can be explained in terms of natural necessity cannot be a matter of choice and therefore cannot be an action for which one is morally responsible. But because Kant claims that one can be free despite being “determined” in accordance with natural laws, he cannot offer this easy answer. Instead, he depends on reaching a kind of reflective equilibrium between his theory of empirical markers and common sense views about moral responsibility. The current challenge is to develop a way to arbitrate between theories when there are conflicting trends within “common sense views.”[40]

[1] Throughout, references to Kant are given using the Academy Edition pagination. For the first Critique, references are to the A and B editions. Where available, I have used translations from the Cambridge Edition of the Works of Immanuel Kant. Translations of the lectures on anthropology (Ak. 25) are my own.

[2] In the first Critique, Kant says something similar: “the real morality of actions . . . remains hidden . . . [because] how much of [our actions] is to be ascribed to mere nature and innocent defects . . ., this no one can discover” (A551/B579n.).

[3] This argument has been widely discussed. Cf. Allison 1990: 214-29, Ameriks 1981, and Beck 1960: 109-25.

[4] Cf. Ameriks 1981.

[5] Within Kantian ethics this has important implication for the scope of moral regard as well, because the sole criterion for moral regard is the capacity for having a good will. Thus if animals can be held morally responsible, then human beings may have direct obligations to them.

[6] The importance of character is discussed in greater detail in Munzel 1999, Kuehn 2001, Jacobs 2003, and Frierson 2003. The empirical nature of character is discussed in Frierson 2003 and Jacobs 2003. This empirical nature as well as character’s connection to the visible spontaneity discussed in Kant’s lectures is discussed in {author’s article}. The latter article in particular shows the connection between character and the higher faculties of cognition and desire, which faculties will be essential to Kant’s account of the empirical markers of moral responsibility.

[7] Arguably, a noumenal Wille is also a necessary condition of moral responsibility, and one might imagine entities who have a noumenal Willkühr without a Wille. If this is possible, then a free Willkühr would be a necessary but not sufficient condition of moral responsibility.

[8] Beck has discussed this ambiguity (and others) with respect to freedom in Beck 1987. See too Allison 1990.

[9] Here again it is important to recognize that Kant’s discussion here is an empirical one. Brian Jacobs puts the point well in the context of discussion the nature of freedom of the higher faculty of desire (the “will”) in Kant’s anthropology: “the ‘arbitrium liberum’ that Kant posits against the animalistic ‘arbirtium brutum’ . . . is a practical empirical concept and one that is observable when a human being resists acting solely according to the ‘pathological’ necessity that characterizes animal will” (Jacobs 2003: 120).

[10] Strictly speaking, he argues that it should be referred to the “philosophical faculty” but only because “the question . . . is purely psychological” (7:213-4). Thus it is clear that the philosophical faculty has jurisdiction here only insofar as it is involved in empirical psychology.

[11] The role of the higher faculty of cognition as a marker of moral responsibility also shows up in Kant’s discussion of character. Character, which is a “sign” (7:285) or “marker” (25: 1156) of freedom, is defined as ““that property of the will by which the subject has tied himself to certain practical principles” (7:292). But “principles” come only from the higher faculty of cognition and thus relate only to the higher faculty of desire. One who lacks a properly functioning understanding cannot formulate action-guiding principles at all, much less “tie himself” to them. Thus the higher faculty of desire, and with it the higher cognitive faculties, become an important precondition of, and thereby marker for, character. The connection between character and the higher faculty of desire is discussed in more detail in {author’s essay}.

[12] There is no reason to see the natural faculty of understanding and the capacity to be moved by pure reason as equivalent. One might have a higher faculty of desire that is immediately determined to act by various principles of the understanding, but those principles might all be ultimately tied to various non-rational desires. Hume’s notion that reason is the slave of the passions represents just such a view, and Kant nowhere claims that the mere fact of intellectual impelling causes shows that purely intellectual motives can move human beings.

[13] This suggestion is confirmed elsewhere. In a late lecture on metaphysics (Vigilantius, 1794-5), Kant introduces moral categories into a discussion of empirical psychology by pointing out that human beings always have a capacity for action from “pure power of choice” (emphasis added) and insists that “a representation . . . of the law of duty is always concurring alongside [any action], . . . because otherwise one would make a human being equal to cattle or the devil” (29:1015). Kant seems to think here that the morally relevant feature of human beings that distinguishes us from animals (and devils) is our capacity for being motivated purely intellectually.

[14] I avoid the term “actions” because some have construed this term to imply moral responsibility analytically.

[15] For more on the nature of affects, see Sorenson 2002, Frierson 2003, and Borges 2004.

[16] The emphasis on the understanding more generally also helps deal with Kant’s reservations about the possibility of empirically determining whether an action is the result of pure reason.

Admittedly, Kant insists that empirical evidence for the psychological features that correlate with moral responsibility will never be decisive, both because of the difficulty of investigating human beings (cf. 7: 121, 131-4, 25: 1212) and because in principle empirical investigation can never yield apodictic certainty (see 4:471, B3, B124/A91). But the limitation on knowledge of freedom in the first Critique is more stringent than these limitations on empirical psychology.

[17] Difficulties with terminology make interpreting Kant here a bit tricky. Kant often uses terms like freedom ambiguously, such that it is not clear whether he refers to transcendental or empirical freedom. If Kant means by “freedom” in this passage only empirical freedom, then he could simply be saying that higher cognitive faculties are insufficient; one must have a higher faculty of desire (hence freedom) in addition. In the context, however, the relevant “freedom” seems to refer to transcendental freedom.

[18] My account of this is necessarily somewhat brief here. I defend in detail the claim that higher faculties of cognitions and desire fit into a series of natural causes in {author’s article}.

[19] In the Groundwork, he points out that “worldly wisdom” involves “the skill of someone in influencing others so as to use them for his own purposes” (4: 416n.). And throughout Kant’s anthropology – both the published Anthropology and his many lectures – Kant emphasizes the importance of anthropological knowledge for influencing the choices of others. Kant bemoans the fact that “morals and preaching that are full of admonitions . . . have little effect” and attributes this failure to “the lack of knowledge of man” (25:471-72, cf. 27:358). In later lectures, Kant again emphasizes the topic of influence on others:

We must trouble ourselves to form the way of thinking . . . of those people with whom we have to do . . . . So we are taught anthropology, which shows us how we can use people to our ends. (25:1436)

In the published Anthropology the same theme emerges. Kant says in his discussion of the different characters of various national groups,

We are interested only in [what] would permit judgment about what each has to know about the other, and how each could use the other to its own advantage. (7:312)

The emphasis in these passages is on the ability to influence the choices and behavior of others.

[20] As he says there,

This happens when the other . . . confronts the subject with . . . the moral law by which he ought to act. If this confrontation makes an impression on the agent, he determines his will by an Idea of reason, creates through his reason that conception of his duty which already lay previously within him, and is only quickened by the other, and determines himself accordingly to the moral law. (27:521)

In his second Critique, Kant goes further and offers details about how one might, through depicting the moral law in a particular way, prompt another person to act from an appreciation for the value of morality. He explains how one can

show in an example the mark by which pure virtue is tested and, representing it as set before, say, a ten-year-old boy for his appraisal, see whether he must necessarily judge so of himself…. One tells him the story of an honest man whom someone wants to induce to join the calumniators of an innocent but otherwise powerless person . . .. Then my young listener will be raised step by step from mere approval to admiration, from that to amazement, and finally to the greatest veneration and a lively wish that he himself could be such a man (though certainly not in such circumstances)…. All the admiration and even the endeavour to resemble this character, here rests wholly on the purity of the moral principle…. Thus morality must have more power over the human heart the more purely it is represented. (5:155-7, my emphasis)

This passage shows how the higher faculty of desire can be shaped by education in very concrete ways, and confirms that even these higher faculties are part of an empirical series of causes and effects.

Kant emphasizes the empirical nature of this potential for motivation by moral reason in the second Critique. He introduces his account of educating the ten-year-old boy by claiming that “We will . . . show, by observations anyone can make, . . . this property of our mind, this receptivity to a pure moral interest and hence the moving force of the pure representation of virtue” (5:152, my emphasis). Just as one presents empirical reasons for believing in various human instincts, propensities, and inclinations, so one can show by observation that human beings have an innate possibility for moral motivation. And the receptivity that one finds in human beings for such motivation is precisely a receptivity to empirical influences, such as the telling of a vivid story of moral virtue. And this receptivity is itself the result of empirical causes, though human science may never be able to discern precisely how it arose from them.

[21] Kant even suggests, in the Anthropology, that the higher mental faculties that correlate with moral responsibility could be reduced, in principle at least, to purely biological explanations. When he argues in favor of making decisions about sanity on psychological rather than medical grounds, his reason is that “physicians and physiologists in general have not reached a deep enough understanding of the mechanical element in man so that they could explain, in terms of it, the seizure that led to the atrocity, or foresee it (without dissecting the body)” (7:214). Here at least, Kant leaves open the possibility that such a “deep enough understanding” might at some point be possible.

[22] For details about the role of politeness in moral cultivation, see Brender 1997 and 1998, Frierson 2003:57-8, and {author’s article #2}.

[23] For discussion of the way in which the beautiful cultivates this love, see Guyer 1993, Allison 2001, and Louden 2000.

[24] For more on the role of unsocial sociability, see Anderson-Gold 2001, Wood 1999, and Wood 1991.

[25] Most of these accounts focus on the gradual development of the natural predispositions, and in particular moral predispositions, in the human species. For more on the nature of this development of moral predispositions, see Frierson 2003: 152-62, Munzel 1999, Anderson-Gold 2001 and 1986, and Wood 1999 and 1991.

[26] It is important not to lose sight of the limits of these empirical accounts. Insofar as the higher faculty of desire is studied within empirical psychology, its choices are to be explained like any other natural event – by reference to natural causes. But human choice cannot ultimately be explained by this natural story. Human choices are also transcendentally free, and in that sense, they are the grounds rather than the effects of any empirical story that might be told. Moreover, any human choice can be thought of from a practical perspective, from which perspective such a choice cannot be considered a mere result of empirical causes. And when one makes a choice on the basis of pure practical reason, such a choice is specifically free from empirical determination – but free only from within the practical perspective.

[27] Needless to say, the fact that Kant claims that transcendental freedom is necessary for moral responsibility is not enough to establish that this claim is correct. For the purpose of articulating an overall picture of Kant’s account of empirical markers, I take this claim for granted here.

[28] Of course, one could still attribute a person’s moral wickedness to him because God would not be the ultimate cause of that wickedness. In that sense, Kant need not remove moral responsibility for the radical evil that makes grace necessary, but he does remove moral responsibility (credit) for the revolution that shifts one from evil to good.

[29] These two cases might not be that different, since God determines the natural laws, at least to a considerable degree. See 5:124-31.

[30] This also means that a skeptic who claims that human beings are in fact causally determined all the way down and thus not morally responsible cannot be refuted by Kant’s defense of human freedom and moral responsibility. As long as the skeptic objects on grounds like the one’s mentioned here, nothing in Kant’s transcendental idealism nor in his empirical psychology can prove that human beings are morally responsible. For more on the “modesty” of Kant’s system, see Ameriks 2000.

[31] Some of the same conclusions about the possibility for moral responsibility for non-rational entities can be drawn from Kant’s account of maxims. A maxim, according to Kant, is a “subjective principle of volition,” (4:400n), a principle on the basis of which one chooses to act. This notion of a maxim, however, is not primarily a concept of empirical psychology, but of practical philosophy. Kant primarily discusses the concept of a maxim as part of general discussions of moral evaluation and deliberation, and only very rarely discusses maxims in the context of his empirical psychology. A maxim is the principle that provides the actual basis of one’s action, whether or not one can discern such a principle by empirical means, including introspection. Given that this ultimate basis of action cannot be experienced, however, there is no necessary reason to preclude even trees from acting on maxims. Of course, it is reasonable to distinguish trees from people on the grounds that only people have a first-person perspective within which talk of maxims makes sense, but even this claim is based on assumptions about what is necessary for a first person perspective, assumptions that Kant’s transcendental idealism helps undermine.

[32] Hume 1740: III.1.1, p. 467.

[33] Kant does distinguish between this character in a general sense and character in the narrower sense appropriate to human moral responsibility. But the account in the first Critique leaves open the possibility of ascribing moral responsibility to any entity with character in the broad sense, even if Kant himself limits responsibility to those with his narrower sense of “character simply” (Character schlechthin, cf. 7:285).

[34] Kant gives some arguments for freedom in the first Critique and in the Groundwork, but for the purposes of this paper, however, I am taking the second Critique to offer Kant’s most mature argument for transcendental freedom. The second Critique does depend on the first, but only to establish the compatibility of freedom with natural necessity, not for any positive arguments for freedom.

The precise details of Kant’s argument in the second Critique are controversial, and I will not defend my particular reading of it here. Broadly speaking, I follow Karl Ameriks in seeing it as a regressive argument, based on the premise that “the moral law . . . is simply given” (Ameriks 1981: 53).

[35] This approach contrasts with Allison, who argues that “the fact is best construed as the consciousness of standing under the moral law and the recognition of this law” (Allison 1990:233). I take this fact to be too psychological. When Kant describes the fact as, for instance, “consciousness of the moral law” (5:31), I take him to refer to the moral law itself as the fact, of which one is conscious. Similarly, one might say, “from the awareness that the sky was getting darker, she concluded that night was approaching.” But no one thinks that one reasoned from “the awareness” to the approach of night. Rather, the relevant premise is “the sky is getting darker.” Likewise when Kant refers to “consciousness of the moral law” as the fact of reason, he is elliptically referring to the moral law itself, or better the fact that there are moral laws for us, that we are obligated to act in certain ways rather than others.

[36] There is another less consequential reason that the argument of the second Critique does not quite apply to saplings. Kant’s argument is addressed to an intelligent practical reasoner. And of course to “judge that one is free,” one must have higher cognitive faculties, because without those, one can judge nothing at all. What this point establishes, however, is only that no sapling can believe that it is free, since no sapling has beliefs. It shows nothing about whether or not a sapling is in fact free.

[37] It will, of course, be more difficult to write an account of “the incentives of pure practical reason” for a tree. But this shows only that the experience of being a tree (if there even is such an experience) is something humans cannot understand. It doesn’t show that the “character” of the tree, the fundamental principles that govern its “activity” in the world, cannot be grounded (dare I say “rooted”) in freedom and even in the moral law.

[38] For a detailed explanation of the role of “inclination” as the governing principle of the lower faculty of desire, see {author’s article}.

[39] Karl Ameriks has helpfully described this approach in Kant as “Kant’s Modest System” (see Ameriks 2000).

[40] {Acknowledgements deleted for anonymity.}