In this article, I discuss an aspect of Kant’s account of moral responsibility that has been recognized as a problem but is still awaiting satisfactory treatment: the question whether Kant’s theory can admit of degrees of responsibility. According to Christine Korsgaard, Kant’s views prima facie imply that we must hold others ‘completely responsible for each and every action, no matter what sorts of pressures they may be under’ (Korsgaard Reference Korsgaard1996: 205). However, as Korsgaard observes, it would be uncompromising to allow for only two possibilities: either full responsibility for everything one does or no responsibility at all, as in the case of small children and animals. A moral theory that does not allow for excuses or mitigating reasons neglects a central feature of our everyday and legal practice. Moreover, the impossibility of excuses would be in tension with Kant’s own statement in the Metaphysics of Morals, where he claims that the ‘degree to which an action can be imputed’ depends on the ‘magnitude of the obstacles that had to be overcome’ (Kant Reference Kant1996a: 19; 6: 228).Footnote 1 My aim is to show that this claim is consistent with Kant’s theory as a whole and thereby to make transparent how degrees of responsibility, and excuses or mitigating reasons in particular, are possible for Kant.
In the first section of this article I provide an account of the concept of responsibility in Kant’s practical philosophy, and I show that Kant frames the topic of responsibility in terms of ‘imputation’: imputing an action to a person means ascribing responsibility to the person for her past action. In the second section I outline the problem of degrees of responsibility as Korsgaard describes it and discuss Kant’s account of transcendental freedom that gives rise to the problem. My own solution takes as its starting point the passage about degrees of imputation in the Metaphysics of Morals, which I analyse in detail in the third section. Then I show that Kant’s account implicitly relies on the distinction between two levels of imputation, which can be found in the legal theory of Kant’s time. My suggestion is to map this distinction onto the distinction between accountability and praise- or blameworthiness as two senses of responsibility, which is employed in current theories of responsibility such as R. J. Wallace’s. In the fifth section I explain that the distinction between the two levels of imputation, or the two senses of responsibility, is crucial for the possibility of degrees of responsibility in Kant’s theory and fits with the general division into an a priori and an empirical part of his moral theory. In the sixth I make use of a typology of excuses given by R. J. Wallace in order to characterize the reasons that can count as excuses or mitigating conditions. Finally, I argue that, whereas praise- or blameworthiness can come in degrees, the other sense of responsibility, accountability, is a threshold-concept not allowing for degrees.Footnote 2
1. Responsibility and Imputation in Kant’s Practical Philosophy
Kant uses the term ‘responsibility’Footnote 3 rarely. Instead, he frames the topic of responsibility for one’s past actions in terms of ‘imputation’. I will therefore start by introducing the concept of imputation and then relate it to the concept of responsibility. Kant defines the concept of imputation in the introduction to the Metaphysics of Morals, in a section titled ‘Preliminary Concepts of the Metaphysics of Morals’, which outlines the fundamental ‘concepts … common to both parts of the Metaphysics of Morals’ (Kant Reference Kant1996a: 15: 6: 222), i.e. to both Kant’s ethics and legal philosophy:
Imputation (imputatio) in the moral sense is the judgment by which someone is regarded as the author (causa libera) of an action, which is then called a deed (factum) and stands under laws.
(Kant Reference Kant1996a: 19; 6: 227)
A judgement of imputation establishes a special relation between an action and ‘someone’ – an agent, i.e. a person (cf. Kant Reference Kant1996a: 16; 6: 223). Imputing an action means regarding the person not only as the natural cause of the action, but as its ‘author’, and this in turn implies that the person is seen as the action’s free cause. An action originating in a free cause is called a ‘deed’ (ibid.).
Furthermore, a judgement of imputation implies that the action ‘stands under laws’. The laws Kant is thinking of here are not laws of nature, but normative laws that are a criterion of an action’s correctness. Kant talks about imputation ‘in the moral sense’, therefore the relevant laws are moral laws, which encompass ethical and legal laws alike (cf. Kant Reference Kant1996a: 14; 6: 214). Moral imputation is thus the generic term for legal and ethical imputation.
Note that the judgement of imputation described so far is not identical with the actual moral evaluation of an action, i.e. the application of a moral law. It is, rather, a necessary condition for the moral evaluation of an action, because it characterizes the action as an appropriate object of moral judgement.Footnote 4 In real life, however, a judgement of imputation is normally motivated by an interest in normatively evaluating the action and the agent. Therefore, it is not surprising that Kant, after he defines imputation, continues with evaluative categories:
If someone does more in the way of duty than he can be constrained by law to do, what he does is meritorious (meritum); if what he does is just exactly what the law requires, he does what is owed (debitum); finally, if what he does is less than the law requires, it is morally culpable (demeritum).
(Kant Reference Kant1996a: 19; 6: 227)
Kant considers three possible results when it comes to moral evaluation: positive, neutral and negative. The positive and negative cases are of most interest for questions of responsibility. Let us first comment on the notion of merit. In the definition, Kant characterizes an action as meritorious if it is more than can be coerced by law. Kant acknowledges two different kinds of coercion or constraint:Footnote 5 ‘external constraint’ and ‘free self-constraint’ (Kant Reference Kant1996a: 148; 6: 383). Since compliance with juridical laws can be externally coerced (ibid.), meritorious actions could be understood as those morally required actions that are not legally coercible.Footnote 6 Consequently, all actions that fulfil ethical duties would be meritorious, since compliance with ethical duties is not externally coercible. However, in the case of narrow ethical duties, like the prohibition against telling lies, the person is ‘coerced by law’ to perform or omit certain actions, where coercion must be understood as free self-constraint. With respect to narrow ethical duties, one cannot do more than that which can be ‘coerced by law’. Only imperfect duties of virtue allow for a ‘playroom’ (Kant Reference Kant1996a: 153; 6: 390) as to which actions to perform, because they only prescribe the adoption of maxims. An action following from such a maxim is not coercible by law in either of the two senses of coercion. Accordingly, Kant says that the ‘[f]ulfilment of them [imperfect duties of virtue] is merit (meritum)=+a’ (ibid.). A paradigmatic Kantian example of an imperfect duty is the duty of beneficence. Consequently, promoting other people’s happiness, e.g. helping my friend’s daughter with her homework, is meritorious.Footnote 7
An action is evaluated negatively, i.e. as culpable, if it is ‘less than the law requires’. Again, the distinction between perfect and imperfect duties is relevant, because the violation of an imperfect duty is not ‘culpability (demeritum)=−a, but rather mere deficiency in moral worth’ (Kant Reference Kant1996a: 153; 6: 390). If I do not help my friend’s daughter with her homework, my action is not culpable, because it was not required. Attributing culpability for an action that fails to realize the end of an ethical duty is only possible indirectly: if one can infer from that action that the person did not adopt the maxim required by the duty, this constitutes culpability. However, Kant holds that it can be difficult to know one’s own maxims, let alone the maxims of others. Consequently, the category of demerit is mainly relevant for juridical duties, and also for narrow ethical duties.
This account of imputation of merit and culpability bears on the notion of responsibility. That an action is imputable to a person is equivalent to saying that the person is responsible for it, at least according to certain notions of responsibility. I restrict the following discussion to two notions that have been used to clarify the concept of responsibility: accountability and praise- or blameworthiness (cf. Wallace Reference Wallace1994). When a person’s actions can be generally imputed to her, she is responsible, in the sense of being accountable. A person is accountable in virtue of certain features or capacities. As we will see in the next section, it is Kant’s view that transcendental freedom makes a person accountable. Praise- or blameworthiness are ascribed to accountable persons in virtue of particular actions that are evaluated with respect to norms.Footnote 8 This means that praise- and blameworthiness require two further necessary conditions besides accountability of the person: first, that the accountable person performed an action that, second, is evaluated with respect to a norm. Regarding the relation between accountability and praise- or blameworthiness, we can here note that accountability is a necessary condition for praise- or blameworthiness, because moral evaluation of actions presupposes the freedom of the acting person.Footnote 9
Kant never identifies merit and culpability with praise- or blameworthiness, but it is plausible to suppose that his account of merit and culpability is at the same time an account of praise- or blameworthiness.Footnote 10 Kant acknowledges praise even for actions that are not meritorious, but only dutiful: in the Groundwork he considers an action done from honour, which nevertheless ‘deserves praise and encouragement’ (Kant Reference Kant1997: 11; 4: 398). If merit is not necessary for praise and even some merely dutiful actions deserve praise, it is plausible to assume that merit is sufficient for praise. That a person is blameworthy for a culpable action hardly needs explanation. In the example of the ‘malicious lie’, Kant holds that ‘one … blames the agent’ (Kant Reference Kant1998: 544; A555/B583) and that blaming is not only something we in fact do, but that is reasonable to do, because it is ‘grounded on the law of reason, which regards reason as a cause that … could have and ought to have determined the conduct of the person to be other than it is’ (ibid.).
Even if Korsgaard regards the notion of praise- and blameworthiness as secondary to the main thrust of her article (Korsgaard Reference Korsgaard1996: 189, 213, n. 5), it is clear that the problem of Kant’s alleged intransigence emerges precisely with respect to these notions. The problem arises ‘when we are making judgments about responsibility: when we must decide whether, for instance, someone is to be exonerated, excused, forgiven, blamed, or not held responsible for a bad action at all’ (Korsgaard Reference Korsgaard1996: 205). The main task, then, consists in explaining how degrees of praise- and blameworthiness are possible in Kant’s theory.
This task becomes less abstract when we consider the practical relevance of these judgements in our moral practice, which for Kant includes legal and ethical contexts. Korsgaard does not discuss responsibility in the legal sphere, but only in personal contexts (Korsgaard Reference Korsgaard1996: 188), and in this article I will not discuss legal responsibility either. However, the relevance of degrees of responsibility in the legal sphere is apparent. Kant considers punishment to be the ‘rightful effect of what is culpable’ (Kant Reference Kant1996a: 19; 6: 227). Therefore a lesser degree of responsibility for a legally culpable action implies a less severe punishment. To grasp the practical relevance of responsibility for a meritorious action, one has to consider the ethical sphere, since, as we have seen, actions that fulfil ethical duties are meritorious.Footnote 11 Kant considers gratitude as the required intersubjective effect of a meritorious action (cf. Kant Reference Kant1996a: 203; 6: 454). If a person renders me a benefit, i.e. fulfils the ethical duty to promote others’ happiness by helping me, I should be grateful. Degrees of merit correspond to degrees in the appropriate response: Kant considers different degrees of gratitude, or of ‘obligation to this virtue’ (Kant Reference Kant1996a: 204; 6: 456), which are proportional to the degree of merit of the other person.
Note that in practice degrees appear to be more important with respect to blame- than to praiseworthiness. It is more important to consider excuses and reduce blame than to determine the degree of praiseworthiness. Nonetheless, there seem to be degrees of praiseworthiness: if a person did something praiseworthy mainly because of some lucky circumstance, we would praise her less than a person who did the same in a difficult situation. One reason for the practical asymmetry of blame- and praiseworthiness might be that we want to avoid or reduce blame and its negative consequences whenever possible, whereas it is not as important to praise a person only to the degree she deserves. I acknowledge this asymmetry between praise- and blameworthiness and will focus more on blameworthiness and excuses.
2. Kant’s Intransigence and its Sources
When it comes to grasping the problem of excuses in Kant’s theory, Korsgaard’s formulation of the worry provides a good starting point:
[W]on’t Kant’s view be intransigent? For if we do regard people as free agents, fellow citizens in the Kingdom of Ends, then it seems as if we must treat them as transcendentally free and so as completely responsible for each and every action, no matter what sorts of pressures they may be under.
(Korsgaard Reference Korsgaard1996: 205)
I take Korsgaard to describe here the problem of degrees of responsibility: the question is whether we should hold persons ‘completely’ responsible for every action, even if they had to face pressures in acting.Footnote 12 I want to stress here that ‘responsible’ is ambiguous. As explained above, there is a sense of responsibility as accountability and a sense of responsibility as praise- or blameworthiness. The question about degrees of responsibility relevant to this article is the question about the possibility of excuses, and excuses function on the level of blameworthiness. Korsgaard refers to this question as well in mentioning excuses. She says, as quoted above, that the problem arises ‘when we are making judgments about responsibility: when we must decide whether, for instance, someone is to be exonerated, excused, forgiven, blamed, or not held responsible for a bad action at all’ (Korsgaard Reference Korsgaard1996: 205; my emphasis).
According to Korsgaard, the feature of Kant’s theory that causes the problem is transcendental freedom, which seems to confer complete responsibility for all our actions. Indeed, Kant sees transcendental freedom – or, equivalently, absolute spontaneity – as the ‘real ground’ of the imputability of our actions (Kant Reference Kant1998: 486; A448/B476). In contrast to merely ‘comparative’ freedom (Kant Reference Kant1996b: 217; 5: 96), transcendental freedom consists not merely in the independence of some empirical circumstances, but absolute independence ‘from all determining causes of the world of the senses’ (Kant Reference Kant1998: 676; A803/B831).
Korsgaard unfolds the problem by considering the solution to the third antinomy. This solution is meant to show how the determination of an event through natural laws is compatible with its being transcendentally free. The key to the solution is the distinction between phenomena, which are governed by natural laws, and noumena, which can enjoy transcendental freedom. Korsgaard rejects an ontological two-world interpretation, according to which there is one world of objects that is causally determined by natural laws and another world of objects that is free. Rather, she favours an account of two standpoints, according to which an object can be regarded as naturally determined from one perspective, but regarded as free from another perspective. When viewing ourselves as phenomena, we occupy the theoretical standpoint and we are concerned with an explanation of our behaviour in terms of natural causes. The practical standpoint allows us to see ourselves as noumenal beings, whose decisions are governed by noumenal laws, and thereby as free and responsible agents who are able to justify our actions by giving reasons. The problem of degrees of responsibility arises when considering the relation between the two standpoints. In the case of excuses and consequent diminished responsibility, we seem to encounter a mixing of the two standpoints, insofar as empirical considerations that belong to the theoretical standpoint enter into a practical judgement about responsibility: ‘The very idea of an action’s being excusable or understandable seems to bring together explanatory and justificatory thoughts. The doctrine of the two standpoints seems to keep such thoughts resolutely apart’ (Korsgaard Reference Korsgaard1996: 206).
The problem becomes even more pressing when we consider a further aspect that Korsgaard neglects. The phenomenal and the noumenal descriptions of the world are for Kant not entirely independent. Rather, Kant claims that an asymmetry holds: if there are free causes, they ground the empirical world, but the empirical world does not exert influence on free causes.Footnote 13 In the solution to the third antinomy, Kant uses the concepts of empirical and intelligible character to refer to the phenomenal and noumenal aspects of causality. With respect to its intelligible character, Kant says, the ‘subject would … have to be declared free of all influences of sensibility and determination by appearances’ (Kant Reference Kant1998: 537; A541/B569), whereas the empirical character belongs to the sensible world and is dependent on the intelligible character, insofar as the empirical character is the ‘mere appearance’ (ibid.) of the intelligible.
If one does not want to follow Kant in the metaphysical assumption of an intelligible ground of the empirical world, the two-aspect-view that Korsgaard (and othersFootnote 14) adopt is a fruitful modification of the Kantian theory. On this basis, the problem of the possibility of excuses in Kant’s theory can be summed up as follows. (1) Agents can be considered both from a noumenal perspective, which presents them as free and responsible, and from an empirical perspective, which presents them as determined through natural laws. (2) These two standpoints cannot be mixed: imputation requires the adoption of the intelligible perspective, i.e. considers the reasons for action, instead of its empirical causes. (3) Excuses point to empirical circumstances (e.g. ‘I acted in anger and therefore could not think about what I was doing’), which can only be acknowledged from the empirical perspective. This, however, requires that one abandon the perspective of imputation. But if this is so, then how can excuses influence a judgement of imputation?
Korsgaard herself advances two proposals intended to attenuate the perceived problem. Her first proposal rests on what she calls Kant’s ‘practical compatibilism’, consisting in the acknowledgement that virtue, i.e. a person’s moral disposition, can be empirically influenced, by education or a just legal constitution, even though it is a moral phenomenon (cf. Korsgaard Reference Korsgaard1996: 210). This is in fact what Kant claims, but it leaves open the question exactly how the analogy is supposed to work. Even if Kant says that empirical factors can influence virtue, it is still an open question how far they can influence a judgement of imputation.
Korsgaard’s second point refers to Kant’s request for ‘generosity of interpretation’: ‘It is, therefore, a duty of virtue … [to] throw the veil of benevolence over their faults, not merely by softening our judgments but also by keeping these judgments to ourselves’ (Kant Reference Kant1996a: 212; 6: 466). It is not clear what Kant means by ‘softening our judgments’, but let us suppose that it means abstaining from judgement and thereby abstaining from imputing the action (fully). Kant might be understood here as criticizing ‘judgemental’ attitudes, i.e. being too ready to attribute fault.Footnote 15 But even if Korsgaard’s proposal is understood to mean that our softening of judgement really reflects the degree of the action’s imputability, the possibility of our judgements being governed by empirical considerations is still in need of explanation. The question how empirical factors can influence judgements of imputation is therefore still open.
3. Kant on Degrees of Imputation in the Metaphysics of Morals
Few recognize that Kant himself acknowledges degrees of responsibility, or, imputation. This is likely because he does so in only one passage of his published works.Footnote 16 Crucially, however, this passage appears right after the definition of the concept of imputation and closes the whole introduction to the Metaphysics of Morals:
Subjectively, the degree to which an action can be imputed (imputabilitas) has to be assessed by the magnitude of the obstacles that had to be overcome. – The greater the natural obstacles (of sensitivity) and the less the moral obstacle (of duty) so much the more merit is to be accounted for a good deed, as when, for example, at considerable self-sacrifice I rescue a complete stranger from great distress.
On the other hand, the less the natural obstacle and the greater the obstacle from grounds of duty, so much the more is a transgression to be imputed (as culpable). – Hence the state of mind of the subject, whether he committed the deed in a state of agitation or with cool deliberation, makes a difference in imputation, which has results. (Kant Reference Kant1996a: 19–20; 6: 228)
Kant expresses a widespread opinion: the greater the effort required to perform a good action, the more we praise the person who performed it. Analogously, we blame a person more if it would have been easy for her to do the right thing. In order to understand the kinds of ‘obstacles’ Kant mentions here, it is helpful to describe in more detail the situation of the acting person Kant alludes to. If a person faces the alternatives of an action that is commanded by duty and an action that is grounded in self-love, she faces obstacles on both sides. The ‘natural obstacles’ are the obstacles to doing one’s duty generated by our sensible nature. All kinds of sensible obstacles are conceivable; Kant picks affects as one example. Those sensible obstacles ultimately exist because the human will is not perfectly rational, but ‘pathologically affected’ (cf. e.g. Kant Reference Kant1996b: 153; 5: 19), i.e. it is not only determined by reason but also affected by sensible motives, which can be opposed to duty.
It is less obvious, at first sight, what a ‘moral obstacle (of duty)’ is. In Kant’s formulation, it seems that for the meritorious act, both natural and moral obstacles must be overcome. This, however, does not make any sense, because only natural obstacles can hinder the morally good deed. Therefore, moral obstacles have to be understood as hindering the omission of the good deed (or, equivalently, the omission of the action incentivized from self-love, which supposedly contradicts duty). According to Kant, every human being is aware of standing under the moral law, and will therefore be reluctant to violate it.
Kant’s formulation – ‘The greater the natural obstacles (of sensitivity) and the less the moral obstacle (of duty) so much the more merit is to be accounted for a good deed’ – must thus be understood as saying that natural obstacles hinder the execution of a good deed, whereas the moral obstacle hinders the omission of the same deed.Footnote 17 If the action in question is culpable, it has to be the moral obstacle that hinders its execution and the natural obstacles that hinder its omission.
Kant claims that there can be degrees of each of the two types of obstacle. It is uncontroversial that there are degrees of natural obstacles that oppose duty, for instance due to differences in the strength of our inclinations. Kant’s account of affects as excuses rests on the assumption that the person for whom strong inclinations motivate a wrong action has to overcome greater obstacles to omitting that action than the person who is in a state that permits ‘cool deliberation’. It is striking that when it comes to both merit and culpability, Kant discusses the influence of obstacles only with respect to there being ‘more’ merit and culpability. However, degrees with respect to there being ‘less’ are logically implied: ‘The less the natural obstacle, the more culpability’ implies ‘the greater the natural obstacle, the less culpability’. This means that affects can figure as excuses for a culpable action.
In contrast to natural hindrances, it is more difficult to see what Kant means by ‘moral hindrances’ and how they can be a matter of degree. I see two possibilities: the moral obstacle could either consist in the obligation itself or in ‘the feelings that accompany the constraining power of the moral law’ (Kant Reference Kant1996a: 165; 6: 406). Both can be understood as being moral hindrances, because they oppose the immoral action, but the second possibility will prove to be more apt for the interpretation of the passage in question.
Regarding the first possibility, obligation as moral hindrance, there seem to be no degrees of duty, since ‘duty and obligation are concepts that express the objective practical necessity of certain actions’ (Kant Reference Kant1996a: 16; 6: 224), and necessity is not a graduated concept. How then could there be degrees of moral obstacles? Kant holds that ‘grounds of obligation (rationes obligandi)’ (ibid.) can be more or less strong. He does not elaborate on his notion of grounds of obligation,Footnote 18 but one can surmise that he takes wide duties to provide grounds of obligation, i.e. moral reasons that are not sufficient for action: Kant acknowledges that one maxim of duty can be limited by another, ‘e.g., love of one’s neighbor in general by love of one’s parents’ (Kant Reference Kant1996a: 153; 6: 390), and this could be taken as an example of the conflict of two grounds of obligation with different strengths. Thus degrees of grounds of obligation can, first, depend on to whom the duty is owed. Second, they might depend on the involved good: a person’s being in deadly peril seems to be a stronger ground of obligation to help than a person’s needing a small favour. This is what Kant means when he says that the ‘degree of obligation’ to gratitude ‘is to be assessed by how useful the favor was’ (Kant Reference Kant1996a: 204; 6: 456). Third, wide duties seem to be weaker grounds than narrow duties (cf. e.g. Theory and Practice 8: 300n.). Kant says in the Vigilantius-lecture that benefiting someone is a far lesser duty than paying one’s debt, and thus the merit of a person who fulfils the duty of beneficence is greater than that of a person who pays her debt (cf. Vigilantius, 27.2,1: 568).
It has been objected that Kant mistakenly considers moral obstacles in the passage from the Metaphysics of Morals, since the passage purports to consider the degree of imputability ‘subjectively’. However, moral obstacles are located on the level of ‘objective’ duty (cf. Joerden Reference Joerden1991). Kant expresses this in one reflection: ‘The magnitude of imputability can be judged objectively according to the degree of obligation or subjectively according to the difficulty’ (Reflexion 6812, 19: 169, my translation and emphasis).Footnote 19 The passage from the Metaphysics of Morals, which considers the degrees of imputability ‘subjectively’, therefore should be read as concerning the ‘difficulty’ a person encountered in acting, and not the degree of obligation.
This leads to the second possibility for understanding the moral obstacle. It can be interpreted as a subjective obstacle by identifying it with the feelings ‘that accompany the constraining power of the moral law’ (Kant Reference Kant1996a: 165; 6: 406). Kant mentions as examples ‘disgust’ and ‘horror’ (ibid.). Those feelings function as moral obstacles, as Kant points out in describing them as making ‘moral aversion sensible’ (ibid.; my emphasis).
In contrast to obstacles on the objective level of duty, feelings belong to the subjective level of sensibility. Kant draws the distinction between ‘subjective’ and ‘objective’ in different ways throughout his works; one way is to identify ‘objective’ with what is the same for every being with reason (i.e. duty, the moral law) and ‘subjective’ with what belongs to sensibility. In this sense, Kant assigns feelings like disgust and horror to an ‘aesthetic of morals’, which serves as a ‘subjective presentation’ of a metaphysics of morals (Kant Reference Kant1996a: 165; 6: 406). Thus the degrees of imputability which Kant discusses in the passage from the Metaphysics of Morals should be seen as ‘subjective’ in the sense that they depend on factors that go back to the sensibility of the person.
In what follows, the natural hindrances will be of more interest, because the central question is how far empirical factors such as natural hindrances can serve as excusing reasons. At any rate, in the Metaphysics of Morals Kant deals with degrees of praise- and blameworthiness, which is what Korsgaard fears Kant could not consistently allow in his theory. In what follows, I want to show how degrees can be integrated into Kant’s overall theory by distinguishing different senses of responsibility and imputation.
4. Two Levels of Imputation
In the passage from the Metaphysics of Morals , Kant speaks of imputation as meritorious or culpable, but not of imputation in general. It is only with respect to imputation in the sense of praise- or blameworthiness that Kant explicitly mentions degrees. This can be taken as a hint as to how degrees of responsibility are possible: different notions or aspects of responsibility might have to be treated differently when it comes to the question whether degrees are possible.
It is helpful to map the distinction between accountability and praise- or blameworthiness onto a distinction between two levels of imputation within the legal theory of Kant’s time.Footnote 20 The first level of imputation, ‘imputation of the deed’ (imputatio facti), corresponds to the above-quoted definition of imputation (Kant Reference Kant1996a: 19; 6: 227). As previously described, this judgement characterizes an action as a possible object of the application of a law, and the person is regarded as the free cause of the deed. This level of imputation, I suggest, is closely related to the sense of responsibility as accountability, since the person is regarded as a free cause and thereby as a person with the capacity for free choice, which is the core capacity of accountable persons. The only difference between the imputation of the deed and the ascription of accountability is that the imputation of the deed already presupposes a deed, whereas accountability refers to a general capacity of the person. Therefore accountability is a condition for the imputation of a deed. As soon as an accountable person performs a deed, it can be imputed to her on the first level.
The next step is the application of a law, i.e. the evaluation of the action with respect to a law. It can in principle yield three different results (cf. section 1): an action can be more than, precisely as much, or less than the law requires. The logical step following the application of the law is the second level of imputation (imputatio legis). This is the judgement about the person’s merit or demerit (the mere fulfilment of duty is not followed by a second level of imputation). Being a judgement about merit and culpability, imputation on the second level is a judgement about the person’s praise- or blameworthiness.
In many judgements of imputation in our moral practices, the distinction between these three steps is not drawn explicitly. Often, we rather simultaneously identify a person as an author of an action and evaluate both the action and the agent. However, the distinction between the two different levels is important, as it allows us to locate degrees of responsibility more precisely. From what has been said so far, it becomes apparent that Kant speaks of degrees of imputability only on the second level of imputation, i.e. with respect to merit or demerit. This already presupposes the first level of imputation as a necessary condition, i.e. it presupposes that the person was the free author of the deed. Of course, this does not in itself exclude the possibility of degrees of accountability (cf. section 7).
5. The Possibility of Degrees of Responsibility
The distinction between two levels of imputation provides the basis for a solution to the problem of degrees of responsibility. On the first level, possession of a transcendentally free will is necessary and sufficient for accountability, i.e. for the imputation of the action to the agent. The imputation of the deed is only possible if we regard the person’s action from a practical, intelligible perspective. In doing so, we occupy a standpoint that differs categorically from the empirical standpoint, which sees the action as a merely natural event. In this respect, Korsgaard is certainly right: it does not make sense to ‘mix’ the two standpoints, as long as we are concerned with the question whether the event is an action (or, in Kant’s terminology: a deed) at all.
Korsgaard supposes that the practical standpoint requires abstracting from all empirical factors, because otherwise this would imply a change in standpoint. However, this would only be the case if, in light of empirical factors, the event did not count as a free action anymore and stood instead as a merely natural event. In this case, one would indeed have switched to the empirical standpoint. Therefore the question is whether all kinds of empirical influence destroy the event’s status as an action. Kant seems to deny that all empirical influences reduce an action to a merely natural event, and it is indeed plausible to suppose that the practical standpoint does not require abstracting from all empirical factors. According to Kant, the practical standpoint has to be given up in favour of the empirical standpoint only if empirical influences undermine the ‘power of reason’ (Kant Reference Kant1998: 545; A556/B584) to the effect that the person cannot guide her action according to her reasoning. The possibility of acting according to one’s reasoning can therefore count as a criterion for whether an empirical influence destroys the status of the event as a free action. When an affect, for example, makes reflection not impossible, but only ‘more difficult’ (Kant Reference Kant1996a: 166; 6: 407), its influence is compatible with regarding the action from a practical standpoint. It is also plausible from a systematic point of view to allow for empirical influence even when judging from a practical standpoint: for human beings, being transcendentally free does not mean acting without any consideration of empirical constraints. Persons exercise their non-empirical capacity in the empirical world and therefore always have to cope with empirical influences. Consequently, on the second level of imputation, precisely those empirical circumstances that do not completely undermine the ‘power of reason’ of the person and thereby the status of the event as a free action can count as excusing reasons.
Having established the possibility of empirical factors as excuses in the Kantian framework, are there any positive reasons for attributing to Kant two levels of imputation and allowing for degrees of responsibility on the second level? I will discuss two such grounds: first, the fit with the overall structure of Kant’s moral theory; second, the role of luck.
First, this two-level structure fits well with the general architecture of Kant’s moral philosophy. According to Kant, a priori and empirical investigations have to be strictly separated. The a priori level undoubtedly has priority because Kant is concerned with the question what duties we have, and those duties are grounded a priori (cf. e.g. Kant Reference Kant1997: 2; 4: 389). The conditions that have to be fulfilled at the first level of imputation – that the person possesses a transcendentally free will – have to be fulfilled in order for persons to have moral duties at all, and thus they also have a certain priority. Still, Kant describes the empirical part of moral philosophy as an indispensable ‘counterpart’ of a metaphysics of morals:
The counterpart of a metaphysics of morals … would be moral anthropology, which, however, would deal only with the subjective conditions in human nature that hinder people or help them in fulfilling the laws of a metaphysics of morals. It would deal with the development, spreading, and strengthening of moral principles … [with] teachings and precepts based on experience. It cannot be dispensed with, but it must not precede a metaphysics of morals or be mixed with it …
(Kant Reference Kant1996a: 10–11; 6: 217, my emphasis)
From this passage we learn that the empirical part of moral philosophy is indispensable when it comes to our having a full picture of moral practice. Kant thinks that a moral philosopher should be interested not only in the grounding of an abstract formal principle, but also in the fulfilment, i.e. the realization, of morality in the empirical world.
It is precisely the successful or deficient execution of moral laws that grounds the imputation of merit and demerit. The two levels of imputation, therefore, reflect Kant’s two-level structure of moral philosophy: the imputation of the deed presupposes the non-empirical capacity to recognize the moral law as an objective principle of action, i.e. autonomy. On the second level of imputation, the realization or appearance of this capacity is relevant, and therefore we are concerned with the question which ‘subjective conditions in human nature’ the person encountered and how successful she was in making the moral law her subjective principle, i.e. her maxim. The indispensability of the empirical level of moral theory is explained by the fact that actual moral agents are empirical agents in the empirical world. Degrees of responsibility or excuses reflect the same insight: in a real moral practice, we do not encounter idealized rational beings, but actual persons who face hindering or helping conditions in the empirical world.
This thought shows that the two standpoints, the noumenal and the empirical, do not have to be ‘mixed’ in order to account for empirical influence on practical judgements and to allow for excuses. Rather, it should be kept in mind that the capacity of reason, which constitutes the noumenal aspect of the human being, has to be realized in the empirical realm. In this activity, the relevance of hindering empirical influences can be considered by letting them count as excuses.
Stephen Engstrom applies this thought to the notion of autonomy. He distinguishes between an objective and a subjective sense of autonomy, the latter referring to the success in making the objective moral law one’s maxim (1988: 439). According to Engstrom, subjective autonomy is conditioned and comes in degrees, depending on the person’s ability to perform the right action in light of obstacles (1988: 449). In contrast to Engstrom, I suggest reserving the notion of autonomy for the capacity to recognize the moral law and using Kant’s own concepts for describing degrees relating to the realization of this capacity in the empirical world: an agent overcoming more or greater obstacles is more praiseworthy or more virtuous. We can now account for the analogy between virtue and responsibility noticed by Korsgaard (cf. section 3). Virtue and praise- or blameworthiness both admit of degrees because both concern the realization of the (non-empirical) capacity to act according to the moral law in the empirical world. Degrees of praise- or blameworthiness accrue to a person because of a particular action, degrees of virtue because of her overall character.
The second reason why degrees of responsibility should play a role in Kant’s moral theory is that acknowledging degrees of responsibility can be understood as a way of dealing with moral luck. Kant does not want to impute to a person what contingently happened to her, but rather what can be traced to her free choice. Therefore a guiding idea for introducing degrees of responsibility could be to concentrate on the aspect of an action that is in the power of the person. Something has to be ‘subtracted’ from culpability if the person encountered natural obstacles, the existence of which she cannot control. For merit, the opposite holds. According to Kant, persons are not responsible for the existence of their inclinations, because these are merely natural phenomena, but for the ‘indulgence’ towards them, i.e. the ‘influence’ on their maxims (Kant Reference Kant1997: 62; 4: 458). If a person is responsible for the influence of her inclinations on her actions, it is also imputable that she resisted the inclinations. If a person performs a good action even in the face of great natural hindrances, she achieves more than a person who – contingently – faces fewer obstacles.Footnote 21 Overcoming natural hindrances shows that human freedom can be realized in the empirical world, and Kant honours the firm resolve of a person who did the right thing despite natural hindrances by designating it as more meritorious.
It is surprising that, even if his theory allows for excuses, Kant hardly considers them. There seems to be a reason: excusing oneself is in tension with the aim to become a morally better person. To be sure, it is possible to take into account that as a human being one is only imperfectly rational and to exercise leniency in excusing oneself for an immoral action. However, to realize the end of improving one’s own ‘moral perfection’ (Kant Reference Kant1996a: 196; 6: 446), it is important to concentrate on one’s own possibilities for rational action, in contrast to considering oneself as helplessly subjected to foreign forces, which could be adduced as excusing reasons. It seems that in many cases one would have had the possibility of omitting the forbidden action. As the gallows example illustrates, Kant holds that it is always possible to act as morally required, even when the natural hindrances are great: even if one’s life could only be saved by telling a lie, it would be possible to speak the truth (cf. Kant Reference Kant1996b: 163; 5: 30). Refraining from excusing oneself can help us to improve as moral persons in a way that avoiding or reducing responsibility by pointing to excusing circumstances cannot. Regarding other people, however, the situation is different, because it is only possible to improve one’s own morality and not the morality of others (cf. Kant Reference Kant1996a: 150; 6: 386). Therefore showing mildness in excusing other people does not conflict with concern for moral perfection.
The distinction between two levels of imputation allows us to see that an excuse can undermine praise- or blameworthiness, while leaving intact the ascription of accountability or the first level of imputation. This means that an action can be imputed to a person who is nonetheless considered blameworthy only to a lower degree, because she can be excused. In the remainder of this article, two more questions are to be answered: what kinds of reason can count as excusing reasons in the Kantian theory, and are there degrees of responsibility at the first level, i.e. degrees of accountability?
6. What is an Excuse?
Before considering excuses in Kant’s theory, it is helpful to characterize excuses in a general way. Excuses do not question the action’s wrongness. This distinguishes them from justifications, which show that, contrary to first impression, the action was not wrong at all.Footnote 22 The structure of imputation can serve to clarify the function of justifications and excuses: justifications concern the application of the law, i.e. the evaluation of the action, because they point to circumstances under which a kind of action that is normally wrong is not wrong at all. Those circumstances constitute exceptions to a prohibition, e.g. ‘To kill another person is wrong, unless it is done in self-defence’. Even though the killing of the other person is imputable to the agent on the first level, i.e. as her own action, the agent is not blameworthy. However, this is not because of an excusing reason, but because of a justification: there is no law that prohibits an act of adequate self-defence.
Excusing a person presupposes that an accountable person performed an action (i.e. the first level of imputation) and there is no justification, i.e. that she did something wrong. The excusing reason concerns the second level of imputation: blameworthiness is reduced.Footnote 23 There is another way of challenging an ascription of responsibility, namely pointing to the fact that a person is not a full member of the moral community, such as a child or a person who is mentally ill. Those exempting Footnote 24 reasons function ‘globally’ insofar as they refer to the agent rather than to some particular act. Those reasons undermine the belief that a person is accountable in the first place, because she does not possess the relevant capacities. We will turn to exempting reasons in the next section.
The aim of this section is to specify what can count as an excuse for Kant. This will be done in two steps. First, by taking a prominent contemporary account of excuses as a comparison, we understand why excuses in Kant’s sense only reduce responsibility instead of undermining it altogether. Second, by contrasting affects and passions, it will become clear that an empirical factor can figure as an excuse if (a) the person is not blameworthy for its existence and (b) it significantly hinders rational deliberation without making it impossible.
A prominent account of excuses in the contemporary debate is that offered by R. J. Wallace. Typical excuses, taken from his typology (cf. Wallace Reference Wallace1994: 136–47), are ignorance (if it is not itself culpable), unintentional bodily movements, physical constraint, and coercion. What all these excuses have in common, according to Wallace, is that they show that the agent did not possess the kind of intention that would make him responsible for the action (cf. Wallace Reference Wallace1994: 128): the person who inadvertently treads on another’s hand did not have the kind of intention (harming someone) that would make her fully blameworthy for the action. Apparently, Kant’s excusing reason – acting from affect – does not fit into any of these categories. A closer look at each of Wallace’s excusing reasons will explain why. First, ignorance means that the person does not know that she is performing an action that is wrong under a different description (treading on someone’s hand) than the one she has in mind (walking to the refrigerator). Her action is not intentional under the description under which it is culpable. An action from affect, however, does not imply lack of knowledge and is performed intentionally under the description under which it is wrong. This shows that Kant is occupied with excuses for actions that are intentional under the given description. Second, Kant is not concerned with unintentional bodily movements, because such movements are not imputable on the first level. If it is not an action at all, responsibility in the sense of blameworthiness is impossible in the first place, and so the question of excuses does not come up. Similar considerations apply to the third case. In a state of physical constraint, a person lacks the possibility of acting – in these circumstances, any omission is not imputable to her at the first level. Only coercion seems to be a case that Kant could consider as an excuse, because coercion still leaves the person some – though limited – room for a decision. It is conceivable that Kant could accept coercion as an ‘outer’ empirical obstacle, which could complement his statements about ‘inner’ natural obstacles like affects in the Metaphysics of Morals.
What we learn from the comparison with Wallace’s list of excuses is that Kant is interested in actions that are imputable to the person as her deed and that are performed intentionally under the same description that presents them as wrong. Whereas in Wallace’s account excuses show that the person did not possess a blameworthy intention and can therefore be fully excused, in Kant’s account it is assumed that the person did in fact act on a blameworthy intention (although only in light of obstacles), and is therefore only partly excused. In order to distinguish the reasons that excuse partly from those reasons that excuse fully, one could call the former ‘mitigating reasons’.
A closer look at Kant’s account of affects discloses more features of excuses or mitigating reasons. At first sight one might surmise that any natural factor could function as excuse. However, in his example of the malicious lie, Kant mentions ‘natural temper’ as an irrelevant factor in the determination of blameworthiness (Kant Reference Kant1998: 544; A554/B582). One possible explanation could be that affects have more motivational force than other natural circumstances that are present in a person’s life more permanently. Natural temper, Kant seems to suppose, forms the background of every rational deliberation and therefore cannot function as an excuse for a particular action. From this, we can conclude that an excuse is a natural factor that has a strong motivating force in a particular situation.
This description not only applies to affects, but also to passions. However, Kant would not accept a passion as an excusing reason. An action out of anger might be excused to a certain degree, but not an action out of hatred. The differences between affects and passions come to light in a passage from the Metaphysics of Morals:
Affects belong to feeling insofar as, preceding reflection, it makes this impossible or more difficult. Hence an affect is called precipitate or rash (animus praeceps), and reason says, through the concept of virtue, that one should get hold of oneself. Yet this weakness in the use of one’s understanding coupled with the strength of one’s emotions is only a lack of virtue and, as it were, something childish and weak, which can indeed coexist with the best will. It even has one good thing about it: that this tempest quickly subsides. Accordingly a propensity to an affect (e.g. anger) does not enter into kinship with vice so readily as does a passion. A passion is a sensible desire that has become a lasting inclination (e.g. hatred, as opposed to anger). The calm with which one gives oneself up to it permits reflection and allows the mind to form principles upon it and so, if inclination lights upon something contrary to the law, to brood upon it, to get it rooted deeply, and so to take up what is evil (as something premeditated) into its maxim. And the evil is then properly evil, that is, a true vice.
(Kant Reference Kant1996a: 166; 6: 408)
The main difference between affects and passions with respect to the possibility of figuring as excuses, I take it, is that affects ‘can indeed coexist with the best will’ and constitute only ‘lack of virtue’, whereas passions are themselves evil (cf. also Anthropology, 7: 267) or at least incentivize evil actions. This difference has two sources: first, a difference in genesis, and second, a difference in the role of deliberation or principles. With regard to their genesis, a person is supposed to take an active part in the formation of a passion, whereas she is a passive victim of an affect. In the above-cited passage, affects are described as a ‘tempest’ that is nearly uncontrollable by the person.Footnote 25 In contrast, the formation of a passion is within the power of the subject. A passion ‘is a sensible desire that has become a lasting inclination’, and this is only possible if situations nourishing the passion occur repeatedly – which is at least partly under one’s control.
The second difference is that an affect makes reflection ‘impossible or more difficult’, whereas a passion ‘allows the mind to form principles upon it’. In the state of passion, a subject embraces a wrong maxim with full premeditation. An affect, in contrast, can tempt a person who has adopted right maxims (i.e. possesses a good will) to perform an action that she will later regret.
This shows a similarity between acting from affect and acting from weakness of the will. Kant characterizes weakness of the will, too, as compatible with a good will (cf. Kant Reference Kant1996c: 84; 6: 37). A weak-willed subject has a morally good maxim, but does not act according to it, because of the presence of tempting inclinations:
I incorporate the good (the law) into the maxim of my power of choice, but this good, which is an irresistible incentive objectively or ideally (in thesi), is subjectively (in hypothesi), the weaker (in comparison with inclination) whenever the maxim is to be followed.
(Kant: Reference Kant1996c: 77; 6: 29)
Whereas actions from affect or weakness go against what the person ‘really’, i.e. on reflection, wants, passions lead to the formation of wrong maxims and thereby to premeditated evil. Because the person, upon reflection, chooses an immoral maxim, she can be said to have a bad will.
Kant discusses weakness of the will in the context of moral, i.e. imputable evil, and therefore one must assume that some maxim underlies even weak-willed actions. For this to be possible, it must be presupposed that the subject possesses contradicting maxims. The same holds for actions from affect: if they are imputable actions on the first level at all, they have to follow from (immoral) maxims of the person, even though those maxims contradict other (moral) maxims the person has adopted upon reflection. In saying that affects make reflection ‘impossible or more difficult’, Kant acknowledges that some affects undermine responsibility altogether, whereas others still allow acting from maxims (those that make reflection not ‘impossible’, but only ‘more difficult’).
To sum up, an excuse in Kant’s sense is a natural factor that is an obstacle to morally good behaviour, as it motivates wrong actions and makes rational deliberation more difficult. To count as an excuse, it should itself not be imputable as culpable to the person (hence passions are excluded from being excuses). Still, the action should be imputable at the first level (that is, deliberation should not be made fully impossible).
7. Accountability as a Threshold-Concept
It is still open whether it is possible for Kant to allow for degrees of imputation on the first level, or degrees of accountability. As indicated above, there are sometimes reasons to exempt someone from being responsible in the sense that accountability is ruled out or undermined. This is the case in early childhood or in the case of insanity, when the human being lacks free power of choice (cf. Lectures on Metaphysics, 28: 255). Making such an exemption has practical consequences: insane criminals, for example, are ‘people whom we should pity and cure, but not punish’ (Kant Reference Kant2006: 108; 7: 214n.).
The possibility of degrees of accountability is much more difficult to integrate into Kant’s theory than degrees of praise- and blameworthiness. The reason why persons are accountable and why actions are imputable on the first level is transcendental freedom, which seems to be an all-or-nothing concept. Therefore, with respect to accountability, or imputability on the first level, Korsgaard’s worry is indeed appropriate.
However, ‘intransigence’ at this level can be defended, or at least be made plausible. Accountability is a central characteristic of persons (cf. Kant Reference Kant1996a: 16; 6: 223), such that not being accountable would at the same time imply the loss of one’s standing as a person. Intransigence with respect to accountability expresses the conviction that the status of personhood, of being a full member of our moral community, is difficult to lose. Even if we experience some ‘pressure’, as Korsgaard calls it, we do not want to see our status as persons easily questioned, and therefore it is plausible to claim that the enjoyment of this status is an all-or-nothing affair. What Korsgaard rightly demands in worrying about degrees of responsibility is recognition that we are neither heroes nor angels – and that we acknowledge our fallible nature by allowing for excuses. Thus degrees are mainly important with regard to blameworthiness, whereas the necessary condition, accountability, could be unaffected.
A serious objection to Kant’s intransigence with respect to accountability is that children are gradually developing capacities of responsible agency. In the case of the mentally ill, too, Kant assumes that there are different degrees of mental impairment (cf. Illnesses of the Head 2: 260, 265). Those degrees of freedom can be described as degrees of the capacity for rational deliberation and action in light of natural obstacles. However, Kant would describe this as empirical freedom. Transcendental freedom, i.e. conceiving of oneself as the first cause of a chain of events, is an idea of reason that cannot be observed empirically. Being the first cause does not admit of degrees: we either assume rational deliberation to be the cause of an action, or we do not. Practical freedom as the capacity to act on rational principles has two aspects: an empirical aspect that can be observed and that admits of degrees, and a transcendental aspect that is an all-or-nothing affair.
This raises the question of how the empirical and the transcendental aspect are connected. How can we conceive of children becoming full persons, i.e. how can increasing empirical freedom lead to the assumption that the person can be seen as transcendentally free? Kant holds that it is ‘impossible to form a concept of the production of a being endowed with freedom through a physical operation’ (Kant Reference Kant1996a: 64; 6: 280). If the genesis of a free person is inconceivable from an empirical point of view, then we cannot even pose the question: when does the free person emerge? The transition from child to full person is beyond any empirical explanation, but the distinction between child and adult could be understood as a normative distinction.Footnote 26 Still, there seems to be an empirical basis for the attribution of a normative status – as Kant says in the first Critique, we must find a ‘ground for thinking of any faculty which is other than sensibly conditioned’ (Kant Reference Kant1998: 540; A546/B575). One plausible suggestion is that, if a certain threshold of empirical freedom, i.e. the capacity to rationally deliberate and to act accordingly, is reached, a subject can be regarded as transcendentally free, i.e. as a full person. In this sense, accountability could be a threshold-concept. Thus, whereas the empirical capacity to act rationally might allow for degrees below and above this threshold, the normative status of a person and, equivalently, accountability, should be understood as an all-or-nothing affair. This view allows for a distinction in kind when it comes to the normative question of who the addressees of duties are, but it allows for degrees on the empirical level.
It has to be noted that there is no metaphysical, unambiguous answer to the question where the empirical threshold of attributing accountability lies. The empirical criteria that serve as a basis for attributing accountability are neither sufficient nor strictly necessary for attributing accountability (cf. Frierson Reference Frierson2008). They are not sufficient, because it is transcendental freedom that is the ‘real ground of … imputability’ (Kant Reference Kant1998: 486; A448/B476). They are not necessary, because the empirical world is underdetermined with respect to the question of whom can be considered transcendentally free. In principle, there is the possibility of attributing transcendental freedom to beings other than human beings or to shift the line between accountable and non-accountable human beings. Kant’s theory allows for the possibility that it is our actual moral practices that determine the empirical criteria for accountability.Footnote 27
8. Conclusion
I have argued that degrees of responsibility are possible in Kant’s theory if two senses of responsibility and, correspondingly, two levels of imputation are distinguished. With regard to accountability and the first level of imputation, Kant’s theory does not admit of degrees. This is due to the fact that transcendental freedom is required in order for one to be an accountable person with a free power of choice and to be the free author of a deed. With regard to praise- or blameworthiness, however, degrees are possible. When accountable persons act, they have to deal with empirical circumstances that shape the exercise of their free choice. These empirical factors are taken into account when considering degrees of praise- or blameworthiness. This solution allows us to answer the most pressing of Korsgaard’s worries: degrees are possible where we need them (with regard to the moral evaluation of an agent), but they are impossible where an all-or-nothing concept is more plausible (with regard to our status as persons).Footnote 28
