The Categorical Imperative is generally stated with two axioms:

  1. Act only according to that maxim whereby you can, at the same time, will that it should become a universal law
  2. Any person who is a means to your end should also be a part of the end.

I find both of these slightly vague, but from some mild interaction with philosophers, I managed to figure that what they really mean is:

  1. If you adopt a certain moral rule, you must adopt it as a reasonable generalisation as the same.
  2. Anyone who contributes something (voluntarily or involuntarily) to your cause must get something out of it that they want.

I get the second axiom (but consider it impractical for reasons outside the scope of this question), but my problem is with the first axiom and the definition of "reasonable".

For instance, if I choose to lie to a tyrannical government on whether I'm hiding fugitives in my house, I can write down the specific moral rule as:

If the Nazis asks you if you're hiding Jews in your house and you are, lie.

When most philosophers say "reasonable generalisation", I think they mean e.g.

If a tyrannical government asks you if you're hiding victims of their oppression in your house and you are, lie.

And not:

Always lie to the government.


Always lie/Never say the truth.


Always commit a given action/Never commit a given action.

Obviously, the last one is self-contradictory, and I doubt most Kantians would support it.

My question is: how much of a reasonable generalisation is a reasonable generalisation? Or am I completely misinterpreting Kant's CI (another possible interpretation that's occured to me is "Do what you wish anyone else did if they were in your situation and you in theirs", which makes much more sense to me)?

2 Answers 2


The main problem behind the categorical imperative is understanding the notion of maxims.

While your intentions may be quite concrete, your maxims are already abstract/general. Maxims of acts, according to Timmermann in "Sittengesetz und Freiheit" (no translation available as far as I know, while some of this may be included in his Cambridge Edition of the Groundwork for Metaphysics of Morals) include a given end and its chosen means. In your example:

Lie in order to help people.

Now this, taken as general law, would mean that it depends on the subjective percievings of a situation (and possible consequences) and helpfulness wether a lie was wrong or right. In his essay On a supposed right to lie because of philantropic concerns he therefore argues that, as no human being can ever be sure of the future and the helpfulness for others, this maxim can never be a general law as it does not really say how to act morally, which for him is linked with certainty.

In the contrary, only not lying can be thought as such a law. Having said this, it depends on the interpreter, but I assume that every maxim that is conditioned by the consequences of the action cannot be morally adequate. The end must be a moral one to do so.

So you essentially did miss the main point of the categorical imperative in your paraphrase.

  • "The end must be a moral one to do so." - But Kant doesn't provide any mechanism to find this moral end. And "no human being can ever be sure of the future" is a silly argument (besides conflicting with the claim that "the end must be a moral one") - you can probabilistically weigh consequences. Commented Nov 22, 2015 at 13:51
  • I think it's this answer that misses the main point of the CI, rather than the question -- Kant states that an action must be chosen such that you'd want a generalisation of the action to be used as a moral rule. Limiting the generalisation to "only moral ends" (i.e. until you no longer like the generalisation) defeats the purpose. Commented Nov 22, 2015 at 13:53
  • @dimension10: I wrote several essays and my bachelor thesis about the CI and no, I do not think I miss the point. In GMM it is clearly stated in Ak. 436 that all three forms of the CI stand for certain aspects of the notion and one form (Formula of Humanity) is explicitly about the choice of ends. If you want a broader analysis of the CI read Allison's commentary of GMM. Although not always accurate, it is quite good in so many aspects.
    – Philip Klöcking
    Commented Nov 22, 2015 at 15:49
  • Does Kant define these ends? Commented Nov 24, 2015 at 15:23
  • He some kind of defines them by his Formula of Humanity: “So act that you use humanity, whether in your own person or in the person of any other, always at the same time as an end, never merely as a means” (Ak. 429). For the possible notions of "end" see SEP. The whole article might be worth a read for you.
    – Philip Klöcking
    Commented Nov 24, 2015 at 15:29

This appears to be a near duplicate of a previous question, though I do not recall the reference.

As Philip Klocking has noted, your paraphrase does not work. However, you are correct that the first categorical imperative (CA) is often viewed as unsatisfactory, described by Hegel and others as an "empty formalism," difficult to apply in practice.

Personally, I find the distinction between "deontic" and "consequentialist" ethics to be most helpful in respect to Kant. In your example, you use what Kant calls a "hypothetical imperative." If you want X, then do Y. The action is guided by the anticipated consequences.

Kant argues that one problem with this in morality is the assumption that we can actually predict such consequences, when Kantian limits on knowledge insist that we cannot. If you hide the liar, then the outcome will be more just. This is unwarranted in itself and we can easily think of many unintended consequences.

So, the CA is much stricter and less commonsensical than we might suppose. We cannot simply think we are doing good. The "universal maxim" must enable the "will" in a way that does not ultimately contradict principles of reason, freedom.... and certain Christian assumption of an immortal soul transcending contingency, nature, and physical constraints, at least that is my understanding.

That said, I'm not sure how to handle your example. Kant does discuss the legitimacy of government, which cannot compel citizens to do what they "may not" do, and this constraint would apply to demands forcing citizens to contradict the CA or, presumably, endanger their immortal souls. So, I believe, your case might possibly be rationalized under the CA, given a closer definition of "tyrannical."

However, Kant is not very lenient towards civil disobedience, and lying is his paradigm example of a CA violation, since lying "universalized" rationally contradicts the very possibility of effectively lying. Your question about the scope of "reasonable generalization" is right on the money. But actually working that out requires working through Kant's entire concept of "reason-freedom" and the CA, and remains, I believe, a vexed question.

However, your case as stated does seem far too "consequentialist" to pass muster with the CA. In practical terms, the CA may be best regarding as a "guiding ideal" by which the internal contradictions of faulty moral assumptions may be revealed.

  • 1
    Can't resist to state that my personal research revealed that Hegel missed the point that maxims have content in form of concrete ends and means and therefore his criticism does not hold, at least in this very form ;)
    – Philip Klöcking
    Commented Nov 21, 2015 at 21:11

You must log in to answer this question.

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