Can Kant tell us what to do? Intentions, dilemmas, and the long journey towards moral perfection

Kant is thoroughly dualistic in his thought, which has gained him some discredit over the years (Meerbote, 1984, p. 57). Sensibility and understanding in his theoretical philosophy, our pathologically affected self and our rational will in his practical philosophy, or the very distinction between theory and practice, are but some of the ravines that cut his philosophy into apparently untraversable regions. We may be prone to criticizing Kant for failing to give a satisfactory resolution to all these various schisms. It might be the case, though, that Kant’s apparent failures say something about the matter itself; some of these tensions may simply be unsolvable.

In this article, I want to focus on the gap between the Kantian ideal of a system of ends perfectly determined by the moral law, and our particular duties before situations of unavoidable harm, arguing that Kantian ethics cannot tell us what to do in cases of inevitable harm if taken as isolate examples, devoid of their due context. In section 1, I argue that, if we interpret such apparent dilemmas as cases of conflicting prima facie duties, Kantian ethics cannot tell us that any of the prima facie duties fails to obligate. In section 2, I try to show that this conclusion is to be expected, since such a framework is foreign to Kantian ethics, and that we should try to approach the issue from the point of view of maxims of action. In section 3, I go on to contend that no maxim that explicitly alludes to harming or allowing harm to human beings can be acceptable under Kantian principles. I conclude in section 4 noting that this requires us to place cases of unavoidable harm in their broader empirical context so that one choice may be reasonably interpreted as “striving” to avoid harm altogether.

1. The shortcomings of a traditional approach

The family of scenarios around the so-called Trolley Problem poses a rather banal conundrum that is, however, quite useful an expositive device. Abstracting from its various derivatives, the core situation is simple: either you save the greater number by killing the lesser number, or you refrain from acting, thereby letting a greater harm be caused. While the reason why the Trolley Problem is a problem at all in current philosophical discussions is not so much what we should do when confronted to it, but rather why our intuitions vary from case to case within the overarching family^[2], I will engage with it from that first point of view, using it as an easily analysable example of cases where, no matter what we do, we cannot avoid all the harm.

This scenario would seem to involve two prima facie duties, and two – and only two – possible actions. We have a prima facie duty (D’.) not to cause death to anyone, and yet another (D’.) to save people in danger, but we can only either (A.) refrain from acting, letting a greater number die, or (A.) act in such a way that will cause the death of the lesser number. A traditional approach to dissolving these problems moves to show that we ought to choose one of the actions (A.) because, given the situation, only one of those prima facie duties succeeds at becoming a duty proper (D.), leaving the other one as merely a prima facie one (Ross, 1987). I aim to show that it is impossible to make this sort of argument either way using Kant’s ethics.

Let us start by applying that argumentative scheme to option A.. Kant’s philosophy seems, at first glance, perfectly suited to do so^[3]. D’. is a perfect duty, since (a) it is a prohibitive duty, with no latitude in its fulfilment (MS 6: 419), and (b) no killing can ever be justifiable, since no rational being can assent to the deliberate destruction of another, innocent rational being. Our humanity has an absolute worth insofar as it is the source of all value (GMS 4: 425) (See Korsgaard, 1996a, pp. 120-121). D’., on the other hand, is an imperfect duty, limited by our capacities at the moment of pursuing it (MS 6: 390). Therefore, it would seem to follow, D’. is constrained by D’.’s prohibitive command. Following Timmermann, we could say that “this wider class of duty […] is based on laws that fail to obligate immediately and without restriction” (2013, p. 46), and claim that, in this scenario where nothing can be done that fulfils both duties, D’. is not binding as a duty proper.

This option, furthermore, looks promising because there are well-known examples where Kant seems to espouse an analogous opinion^[4], and the difference between “strict” and “wide” duties would seem to give rise to a natural interpretation of Kant’s otherwise very rare explicit mentions of moral conflict: it is the stronger groundthat determines what to do when duties apparently collide (fortior ratio obligandi vincit) (MS 6: 224). What else could that mean?^[5]

Nevertheless, we cannot merely base D’.’s overriding status on the terminological distinction between perfect and imperfect duties. It may be helpful to reconstruct Kant’s derivation of imperfect duties to clarify the sense in which they are binding to us in the first place:

P1. A practical law is a determination of the will such that all rational beings recognize it as valid. (KpV 5: 19)

P2. Since we are imperfectly rational beings, we recognize the validity of practical laws, but we do not necessarily follow them – they are our duties. (KpV 5: 32)

C1. Therefore, our duties amount to determining our wills in such a way that any rational being can recognize as valid, that is, to make the maxim of our actions such that it may be willed as a universal law at the same time (from P1 and P2). (KpV 5: 30, GMS 4: 425)

P3. Wills are fundamentally purposive, that is, all maxims of action are end-directed. (MS 6; 385) (see Wood, 1999, p. 51)

C2. Therefore, there will be some ends we have the duty to strive for. (from C1 and P3)

P4. However, an end cannot be the ground for our duties since the only ground that can be regarded as valid by all rational beings is a formal one. (KpV 5: 27)

C3. Therefore, it is our duty to strive for some ends, but whether we achieve them does not determine whether we have fulfilled our duty. (from C2 and P4)

We call these duties imperfect, because there is room for latitude in their material fulfilment (that is, in actually producing the ends it establishes as our duties). There is latitude in the fulfilment of our imperfect duties (MS 6: 390), which immediately follows from the fact that our capacities for action are always finite and limited, but that does not mean that our wide duty may ever “fail to obligate”. Our obligation regarding the objective ends of reason, by itself, can never be overridden or put aside, regardless of how restrictive the situation is for the ends to be actually realized.

This fact about Kant’s ethics muddles the initial framework from which we had said to derive our course of action. While D’. may still be an obstacle to the consecution of D’., this says nothing about D’. being binding or not, since their grounds of justification are not heterogeneous or lexically ordered. What should we do, then, about D’. if we cannot say it is “overridden”?

Some have argued (Cummiskey, 1990)^[6] that, actually, it is A. that is warranted by a closer reading of Kant’s doctrine. If the single ground that justifies ethics in general is universal validity for all rational beings, which, furthermore, are the objective ends of our actions – Cummiskey argues – then the lesser number that will need to die are not entitled to preferential treatment (1990, p. 599). The objective end of moral action, he claims, is the existence of rational beings (1990, p. 606). Thus, D’. fails to obligate in this scenario because the actual root of all our duties is the furtherance and preservation of humanity, which, in some unfortunate cases, requires making sacrifices (1990, p. 603).

This cannot be an acceptable solution either. Cummiskey gets his priorities backwards by starting from the value of rational beings as though they were “natural sources” of value. His claim that “Kant’s normative theory […] is based on an unconditionally valuable, objective end” (1990, p. 597) is right if he means that value depends on rational beings as end-pursuers, but it cannot be sustained if he means that the correctness of our actions is measured immediately with regards to these ends. Kant explicitly denies that very possibility (KpV 5: 21-2) (see Gregor, 1962, p. 76; Wood, 1999, p. 112.). What is good is what every rational being can accept as such, which in turn makes promoting rational beings our unconditional duty, but not the other way around. D’. and D’. cannot be meshed together into a general duty towards the “existence of rational beings”, or, at least, it cannot be so meshed in a way that they may merely be seen as quantitatively distinct forms of the same particular duty. If D’. is to be overridden, that will be because any rational agent would agree to it under those circumstances, not because a larger amount of rational beings will come out of it. However controversial the notion of universal validity for rational beings may be, it does not establish the value of rational beings as “morally relevant objects”, and less so as fungible objects.

2. The general problem

So far, arguing for either A. or A. from a Kantian perspective is looking quite difficult. Of course, noting that several approaches are insufficient to yield a satisfactory answer is nowhere near proving that it cannot be done. Such a proof, strictly speaking, would be woefully hard to attain. Nevertheless, there are two general conclusions to be extracted from the discussion so far.

(1) The framework of prima facie duties is decidedly non-Kantian.

(2) Analysing the conflict in terms of actions^[7] is decidedly non-Kantian.

The main source of puzzlement throughout last section, as it may be apparent, was the particular way of framing the problem. It has seemed impossible to argue that any of our prima facie duties may, in and of itself, “give way” to the other one, but that is to be expected. If we grant that D’. and D’. are, in any sense, Kantian duties, then they express practical necessity (GMS 4: 414-5, KpV 5: 20, MS 6: 222), which means that, either they are genuine and proper, or they are not duties at all. Some may be sceptical about this claim, but the point can be pressed by moving to conclusion (2). That analysing actions is a non-Kantian way of proceeding should not be too controversial (Gregor, 1962, p. 70; Harrison, 1968, p. 299). Kant, indeed, asserts that “it is only the determination of the will and the ground of determination of its maxim […] that matters here [in a critique of practical reason], not its success” (KpV 5: 46). It is maxims, determinations of the will, that form the core of Kantian ethics, as can be plainly seen in the first and third formulations of the categorical imperative (GMS 4: 421, 434)^[8]. To use Esser’s on-point expression, “actions are always subjects of the ethical judging as manifestations of determinations of the will only” (2008, p. 286).

The focus of Kantian ethics on maxims makes it clearer that an analysis in terms of prima facie duties cannot be right. Whether a maxim can be valid (gilt) as universal law cannot be a context-dependent matter. This is a claim that needs qualification, however, since it is evident that maxims usually refer to particular contexts (Harrison, 1968, p. 231). Maxims may incorporate particular contexts that make them acceptable or not. Compare “I shall work as a cook to earn a living” with that same maxim, except working as a cook for the Nazis. Once it is determined that a maxim, with its context of applicability included, is valid, that fact is not context-dependent, and, therefore, its bindingness cannot be subject to qualification^[9].

In conclusion, while there may well be some way to construe Kant’s ethics following the actions-prima facie duties framework that could resolve the issue, it is now clear why it should not shock us if that proves to be a fruitless task, since those are categories pretty much absent from Kant’s approach to judging our behaviour.

3. Maxims of a special kind

Can we, then, use the more Kantian method of analysing maxims to determine what is the right course of action in this case? Let S be the situation such that the agent in question can only do either A. or A.. We may, then, define two maxims:

M.: “Under the circumstances of S, I will refrain from saving the greater number in order to not cause the death of the lesser number”

M.: “Under the circumstances of S, I will save the greater number, even if it means causing the death of the lesser number”

Let us test these maxims first according to the Formula of Universal Law (FUL). The maxims are, quite clearly, not contradictory in conception^[10]. Regarding whether either of them are contradictory in volition, however, a quite poignant problem arises. Whether we can will a maxim depends on how that would fare with whatever we, as rational agents, must necessarily will at the same time (Wood, 2017, p. 41)^[11]. As I discussed earlier, rational agents are, foremost, end-pursuers (Korsgaard, 1996b, pp. 110-111). Thus, we may not will anything that would interdict the conditions for end-pursuing. That is the basis of our duty to not intoxicate ourselves to the point of harming our rational capabilities, for instance (MS 6: 427).

This is essentially the same as the negative dimension (see Wood, 1999, p. 113) of FH. If there is something we cannot will that it become a universal law, granted that it “passes” the contradiction in conception test (CCT), that means that, in some sense, it entails the hindrance of our humanity or that of others. Thus, we can extend Korsgaard’s characterization of FH to the contradiction in volition test (CVT): “as an unconditional end, [the capacity for rational choice] must never be acted against. It is not one end among others, to be weighed along with the rest” (Wood, 1999, p. 125).

CVT and FH, then, would seem to categorically prohibit M.. Now, do they allow for M.? Consider a situation S’ where the agent is faced with the exclusive choice of killing either one person or five people, and the following maxim:

M.: “Under the circumstances of S’, I will kill the greater number in order to not cause the death of the lesser number”.

Needless to say, this is a horrifying situation, and M. is evidently wrong as per CVT and FH. However, as we have already seen, we cannot simply claim that negative duties trump positive ones. If there is no ground to favour negative duties over positive duties, and if we cannot will any maxim that calls for the destruction or hindrance of rational beings, then we cannot will any maxim that explicitly allows it either. Willing to refrain from saving others is an explicit forfeiture of a duty towards others^[12].

We may want to try and circumvent this restriction by coming up with maxims that do not mention this unfortunate state of affairs. Of course, an obvious restriction is in place. We may not come up with permissible yet spurious maxims to bypass our duty. When the lives of people are in danger, it is not the time to contemplate the beautiful in nature, for example. Kant realizes this, adding to his ethical thought the idea of “a self-judging capacity of moral judgment” (RGV 6: 186), showing that a “good will”, understood as a merely subjective state of self-satisfaction with one’s maxims, is not enough for morality; there needs to be an active exercise in reconstructing our actions and determining whether we were actually acting according to and from our duty (MS 6: 446-7) (see Dietrichson, 1968, p. 324). All in all, it is quite clear that a meta-maxim to adopt spurious maxims cannot be effected from a motive of duty.

However, a maxim could avoid willing the destruction of other rational beings while not being spurious to the circumstances. That is the crux of Kleingeld’s argument to use Kant’s FH as a way to account for our different attitudes towards the variations of the Trolley Problem. She provides the following maxim as an example:

When forced to choose in S, I will save more rather than fewer human lives, provided I do not use anyone as a means to this end without their actual consent. Now I am forced to choose in S: either I let the trolley continue towards the five, which will save one life, or I divert it towards the heavy man, which will kill him but save five lives. Five is more than one, so I shall save the five by diverting the trolley (Kleingeld, 2020, p. 220).

Of course, for this maxim to be valid, we must assume from the beginning that “I divert it towards the heavy man” is not using him as a means, which is not clear at all. This follows from the general assumption that, in the standard Bystander scenario, steering the trolley does not use anyone as a means: “[i]n fact, she could save even more lives if he were not there” (Kleingeld, 2020, p. 216). But if our morally relevant attitude towards that person comes from our maxims, the question we should ask is: what role do they exactly play in our practical reasoning, if it is not as means? Our maxim has to include the lesser number who are to die, since it is precisely their being fewer people that justifies steering the trolley away. If that is so, the agent’s practical reasoning is of the form:

(1) I must minimize the number of people who die.

(2) In order to minimize the number of people who die, I must cause the death of the lesser number.

(3) Therefore, I must cause the death of the lesser number.

Premise (2) is clearly instrumental in its form; causing the death of the lesser number is a means to achieving, under said circumstances, the minimal harm possible. And the fact that fewer people are being killed is what sells the deal, the maxim does not work if we ignore the fact that the particular people who are on the second track have been so unlucky as to group themselves with fewer companions – if that track had been occupied by more people than the other one, the agent would not have steered.

Since maxims of action that are relevant to trolley-like problems cannot circumvent the fact that there are people in both tracks, and since, we have seen, no acceptable maxim can explicitly call for or allow anyone’s death, there is only one maxim that seems both acceptable and relevant: “I shall strive to save everyone”. Thus, we arrive at the core of the Kantian response to situations of unavoidable harm. Even though the circumstances may make it unattainable, the only maxim that is acceptable for a rational being in general is that which makes no exceptions among people; the moment the possibility of killing others entersthe maxim instead of being an unfortunate consequence of acceptable maxims, it seems like those people, either are being used as mere means, or, in any case, are not being respected as their absolute worth as human beings would require^[13].

4. The ideal and the mundane

The charge of excessive perfectionism looms sharply over this conclusion. If the only thing Kant can tell us is that we should strive to do what, by the very premise of the dilemma, we cannot attain, how can we ever pretend to act on Kant’s doctrine? Furthermore, it seems like we are violating a revered principle of Kant’s ethical thought: ought implies can. How can it be the case that the only maxims we can adopt in this scenario are ones that command us to do something impossible?

Kant’s actual wording of this principle gives us a crucial clue as to how Kantian ethics is supposed to apply in particular scenarios: “it would not be a duty, to pursue certain realization of our will, if it were not possible in our experience (be it completely, or such that its completion may be thought as indefinitely approachable)” (TP 8: 276-7). This parenthetical clarification is crucial in order to understand Kant’s approach to apparent dilemmas. When pursuing our duty, we can determine our will according to a good maxim, and we can progressively approach whatever ends are compulsory for us.

But what should we do, then? What does it meanto “approach” the realization of our will in this case? We should fend off any temptations to answer in numerical terms. Saving the greater number may be the way to approach saving all of them, but not because the number of people alive will be closer to the ideal^[14]. The absolute worth of all human beings entails that there is a qualitative difference between everyone being saved and anyone dying. Kant cannot solve trolley-like scenarios without additional information. It is only within the grander empirical context of our actual lives that “striving” to save the greater number makes sense^[15]. We may steer the trolley in hopes that the greater number will realize and try to warn the other person; we may not do so and try to call someone who can help^[16]. Ultimately, we may fail, but having taken into account the broader context of that situation, there is a chance for us to actually stay loyal to the very tough demands of morality.

5. Conclusion

I have tried to show that isolate dilemmas cannot be solved under Kantian ethics because they fail to provide us with the necessary context to make the best possible choice given our unfettered obligation towards each and every person. Under Kantian ethics, our duties towards others, insofar as they are duties, cannot fail to obligate; situations of unavoidable harm are always somehow wrong. In order to have any criterion to act under such circumstances, we must have the necessary context to meaningfully say that we are strivingto achieve an end which, at the time of our acting, is unattainable.

Bibliography

Akhlaghi, F. (2020). On Moral Obligations and Our Chances of Fulfilling Them. Ethical Theory and Moral Practice, 23, 625-638.

Cholbi, M. (2009). The Murderer at the Door: What Kant Should Have Said. Philosophy and Phenomenological Research, 79(1), 17-46.

Cummiskey, D. (1990). Kantian Consequentialism. Ethics, 100(3), 586-615.

Dietrichson, P. (1968). What Does Kant Mean by ‘Acting from Duty’? In R. P. Wolff (Ed.), Kant. A collection of critical essays (pp. 314-331). Macmillan and Co.

Donagan, A. (1987). Consistency in Rationalist Moral Systems. In C. W. Gowans (Ed.), Moral Dilemmas (pp. 271-291). OUP.

Esser, A. (2008). Kant on Solving Moral Conflicts. In M. Betzler (Ed.), Kant’s Ethics of Virtue (pp. 279-302). de Gruyter.

Foot, P. (1967). The Problem of Abortion and the Doctrine of the Double Effect. Oxford Review, 5, 5-15.

Gregor, M. J. (1962). Laws of Freedom. A Study of Kant’s Method of Applying the Categorical Imperative in the Metaphysik der Sitten. Barnes & Noble.

Harrison, J. (1968). Kant’s Examples of the First Formulation of the Categorical Imperative. In R. P. Wolff (Ed.), Kant. A collection of critical essays (pp. 228-246). Macmillan and Co.

Kleingeld, P. (2018). A Contradiction of the Right Kind: Convenience Killing and Kant’s Formula of Universal Law. The Philosophical Quarterly, 274(69), 64-81.

Kleingeld, P. (2020). A Kantian solution to the Trolley Problem. Oxford Studies in Normative Ethics, 10(2020), 204-228.

Korsgaard, C. (1996b). Creating the kingdom of ends. CUP.

Marcus, R. B. (1987). Moral Dilemmas and Consistency. In C. Gowans (Ed.), Moral Dilemmas(pp. 188-205). OUP.

Meerbote, R. (1984). Commentary: Kant on Freedom and the Rational and Morally Good Will. In A. W. Wood (Ed.), Self and Nature in Kant’s Philosophy (pp. 57-73). Cornell University Press.

Ross, D. (1987). Prima Facie Duties. In C. Gowans (Ed.), Moral Dilemmas (pp. 83-101). OUP.

Thomson, J. J. (1985). The Trolley Problem. The Yale Law Journal, 94, 1395-1415.

Thomson, J. J. (2008). Turning the trolley. Philosophy & Public Affairs, 36(4), 359-374.

Timmermann, J. (2013). Kantian Dilemmas? Moral Conflict in Kant’s Ethical Theory. Archiv für Geschichte der Philosophie, 95(1), 36-64.

Von Wright, G. H. (1963). Norm and Action. A Logical Enquiry. Routledge & Kegan Paul.

Wood, A. W. (1999). Kant’s Ethical Thought. CUP.

Wood, A. W. (2017). Formulas of the Moral Law. Cambridge Elements.

Notes