Many formulations of utilitarian consequentialism famously lead to a range of "repugnant conclusions", such as:

  • it would be moral to execute an innocent person if this act could deter at least two future murders
  • the Mere Addition paradox, that a world containing the largest number of people whose lives were just marginally better than insufferable has a high amount of total utility
  • the Utility Monster

etc. There is a famous "repugnant conclusion" following from Kant's well-known categorical imperative in the following form:

Act only according to that maxim by which you can at the same time will that it should become a universal law

From this follows that if the Gestapo shows up at one's door, one should not lie when they ask if one is hiding a wanted Jew.

Kant proposed other formulations of the categorical imperative, such as the Humanity Formulation:

The practical imperative, therefore, is the following: Act so that you treat humanity, whether in your own person or in that of another, always as an end and never as a means only.

Now to Kant, these were seemingly interchangeable, and the categorical rejection of lying could be derived from this second form. But at first glance, it doesn't seem that e.g. the imperative to never only treat a person as a means to and end, but always as an end in themselves would automatically lead one to give up a fugitive just so one doesn't have to lie, particularly when considering side effects (see e.g. this).

What are the canonical counterexamples brought forward against Kant's Categorical Imperative in the form of the Humanity Formulation, that one should always treat people as ends in themselves, never merely as means to an end?

  • books.google.fr/… interesting that you call it "repugnant" rather than regretful or something. bernard williams discusses some of the literature on this in truth and truthfulness, iirc
    – user6917
    Sep 2, 2016 at 20:21
  • 4
    I didn't call it that, that's the official name: plato.stanford.edu/entries/repugnant-conclusion
    – jona
    Sep 2, 2016 at 20:49
  • 2
    Your title is missing an important only. There's nothing with using people as means to your ends for Kant -- so long as that's not the only thing going on in an interaction.
    – virmaior
    Sep 3, 2016 at 4:31
  • There's several objections to Kant, but few take on the canonical status of the repugnant conclusion... Most of them focus on universality. I'm not immediately aware of any that focus on never treat people as mere means but always as ends.
    – virmaior
    Sep 3, 2016 at 4:33
  • @virmaior is this lack of canonical objections somewhere discussed in the literature? I have also added the "mere" to the title.
    – jona
    Sep 3, 2016 at 18:24

3 Answers 3


The objections Isaacson lists are important, but the canonical counterexamples are sacrifical dilemmas: Essentialy these scenarios present a situation in which killing a small amount of people can save a great amount of people from death. An examples for these scenarios are trolley-cases.

Every Kantian has to deal with these kinds of problems and other than Isaacson suggests, most of them bite the bullet. E.g. in Germany, whose constitution is highly influenced by Kant, the suppreme court ruled that it is illeagal to shoot down a passenger plane captured by terrorists, even if it would be clear that they want to destroy skyscrapers/ a stadium/ a nuclear power plant.

  • Interestingly, by the ruling of the German Supreme Court (Bundesverfassungsgericht), you hit the nail on the head regarding why these objections do not bother Kant: When you have the decision between two immoral opportunities, how is a principle of morality supposed to be of any practical help? It is not a practical rule that helps to decide what to do in every situation, just whether it's morally permissable to act a given way. That's exactly why he so carefully distinguishes between morals and ethics. And why in morals, there are no dilemmata (see 6:224).
    – Philip Klöcking
    Feb 11, 2017 at 14:24
  • @Philip So how do you know that your two options are "immoral" prior to applying a supposedly useful principle of morality? This is the nature of Mill's and Schopenhaur's objection. In order to use the principle only in the right place, you must already know both the likely consequences of both conflicting choices and have judged both to be immoral. Possessed of both these things, you are able to make moral choices without further rules being required.
    – user22791
    Feb 11, 2017 at 16:41
  • @Isaacson: The objection raised by moral dilemmata is exactly that the CI does say "do nothing", i.e. in application, there is no moral alternative. And this is mistaken imho. Schopenhauer's and Mill's objection is wrong out of other reasons, see this answer of mine (last point b)). They - in my understanding - misunderstood how universalisation works according to Kant
    – Philip Klöcking
    Feb 11, 2017 at 18:17
  • @Philip I don't understand the distinction you're trying to make (in the linked answer) between estimating the actual consequences and estimating what a society of rational individuals would be like. For Mill, at least, it's not the nature of the prediction that bothers him, its the rational power with which it is carried out. It seems excessive for Kant to imply that examining the consequences of our action is fraught with too much potential error, but imagining what a world of rational beings all following our maxim would be like is well within our grasp.
    – user22791
    Feb 12, 2017 at 7:45
  • But more to the point of this actual answer, the fact that those claiming to follow Kant actually do make moral choices in cases that appear to be counterexamples to the successful application of CI formulations does not have any bearing on Kant's formulation itself unless they have applied it as he intended, which this answer does not yet demonstrate. I would be improved if @paschep could demonstrate the steps that the supreme court might have taken to reach their conclusion using Kant's formulations. As it stands, a consequentialist might have reached the same conclusion.
    – user22791
    Feb 12, 2017 at 7:49

Most canonical criticisms of Kant attack the first formulation - Act only according to that maxim by which you can at the same time will that it should become a universal law. Since Kant claims directly that all formulations are essentially rephrasings of one single rule, criticisms of one can be taken to be criticisms of the others.

Hegel's argument touched on the humanity formulation only in that if we all shared the same ends it could lead to conclusion which would be repugnant given that this is not the state of affairs which currently exists. For example, that there should be no private property is not immediately a problem if all people consider this a reasonable circumstance. No-one would take the items you were making use of because you would simply take them back, the situation would continue ad infinitum. Because some people do believe in private property, however, and would defend it with violence if necessary, we must all defend private property in order to maintain our dignity.

Schopenhauer criticises the formulation as reducing only to egoism. His argument is that if we deny the feelings (one of which is empathy), then all we are left with as means of judging the ends of others is our own egoistic ends (Schopenhauer, as I mentioned above, actually phrases this in the context of the universalisation formulation, I've put it in terms of the humanistic formulation to best answer your question). Essentially for Schopenhauer, Kant does not do an adequate job of describing what constitutes permissible ends without requiring further formulations which end up utilitarian.

The most compelling criticism comes from Mill who argues (from a different perspective to Schopenhauer), that Kant's ethics ends up utilitarian in any case. In order to answer the question of whether the maxim could be universalised, we are asking ourselves what the consequences of it would be. If we are capable of determining , with some accuracy, the consequences of global adoption of some behaviour then Kant's objection to consequentialist ethics (that consequences are difficult to judge accurately) either become irrelevant, or undermines his own method.

The efforts of neo-Kantians (and Kant himself) to show how certain scenarios could be worked out using Kantian ethics even though they at first seem to be counterexamples is in fact a demonstration of Schopenhaur's and Mill's criticisms. If a theory does not lead us to counter-intuitive conclusions then is is useless (other than in a descriptive capacity), we obviously already know right actions from wrong ones, we know that letting the Nazi in and telling them where the Jewish family is hiding is wrong. If we know this already such that we need to write caveats and re-interpretations of Kant's ethics in order to make the conclusions fit, then the system is just utilitarianism or evolutionary ethics with the added inconvenience of having to post hoc describe how one use the categorical imperative correctly to arrive at the action one knew was right in the first place.


First the example you give is false: the only two options are not to lie or to give up the fugitive.

There are a range of other options. One among them is to refuse to comply with the process by not answering and blocking attempts at unwarranted searches, whether or not you are hiding someone.

It can be universalized: if people in general refused to allow unwarranted searches, the Gestapo would still not know who was hiding people.

The second form leads to this -- you should resist being a means, without reducing those asking you to mere means. If you think of them as people, not as means in use by others, then you would not bother lying to them.

Kant wrote a very explicit paper about the objections that his position against lying had these kinds of political effects. One interpretation of his position is that open non-compliance is always more moral than covert non-compliance in any such situation, and choosing the latter is a form of cowardice which ends up sacrificing your more principled allies as mere means.

The primary weakness pointed out in the means-ends formulation is that it says nothing in the face of two individuals whose 'ends-ness' must be chosen between. (But that is why he carefully laid out the other equivalent formulations. The idea is that one or the other might be a better starting point, but that they are all, in fact logically identical in implication. If they appear to conflict, your maxim is malformed.)

To my mind, he is particularly evasive on the issue of war. He comes down clearly as anti-war (over and over again) but still avoids stating that declaring war itself is inherently immoral.

The two populations whose ends-ness are at odds, in the issue of war, are never the two fighting sides, but the civilian and military populations.

It is true that soldiers can accept their position and can adopt an honorable stance toward one another that would let them accept death at the hands of a rival faction when their dedication to their own principles are very strong, and no other resolution seems possible. So they can be more than mere means, and there theoretically could be a just war.

Historically they just never have been so. No officer can actually abide the dictum: all but a very few soldiers under his command are playing their part as mere means, the existence of a minority who can see the situation clearly and accept it notwithstanding. Unfortunately, to my mind, the ethics itself is too abstract and too evenhanded to capture this kind of intrinsic inequality of composition.

The construction does not allow for something to be immoral simply because of the real limitations of most human beings. If 90% of us are dupes, still not all of us are dupes. The duped soldiers' willingness to be there, and to make the promises they have made, is theoretically their own business; questioning it betrays their autonomy.

The formulation prevents him from being clear here, and stating the ethical position that clearly represents his own political position.

  • The problem is that the act of hiding is itself a lie. You are doing something while pretending you are not. And, of course, it is very doubtful that we can will that everyone hides people in their basements, in an abstract way that does not takes into account why are we hiding them. Sep 3, 2016 at 11:01
  • Same here: I am not asking for a discussion of the Categorical Imperative and its implications for lying, but: What are the canonical counterexamples brought forward against Kant's Categorical Imperative in the form of the Humanity Formulation?
    – jona
    Sep 3, 2016 at 18:24
  • Your example is still false, nothing in Kant suggests handing over the fugitive. He carefully defended the fact that his position against lying does not lead to this interpretation. And no, law is not ethics, so hiding someone from unjust treatment by the State is not lying. You have not been asked until you have been asked. Most of these 'Canonical examples' are similarly putting words in Kant's mouth that he carefully avoided saying.
    – user9166
    Sep 5, 2016 at 11:29
  • So I have left the objection to the example, but added an issue that is legitimately brought up, and handled poorly by the rule.
    – user9166
    Sep 5, 2016 at 11:43
  • "nothing in Kant suggests handing over the fugitive" I didn't claim he did. I said: "one should not lie when they ask if one is hiding a wanted [person]". Furthermore, I still do not think this is an answer to my question. Now it's just a longer non-answer. I'm not asking for an explanation or defense of Kant's ethics, but for canonical counterexamples brought forward against Kant's Categorical Imperative in the form of the Humanity Formulation.
    – jona
    Sep 5, 2016 at 14:19

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .