Kant argues that one should only act according to those principles which can be universalised. However human actions are often complex and motivated by multi-layered reasonings. They can rarely be distilled down to a simple one-sentence principle.

So when Kant does just that, how is he not committing a fallacy?

For example I may reason that it is ok to lie to a murderer however Kant would argue that I believe it is ok to lie in general and use that as my principle in order to drive a contradiction But I do not condone lying in general, so how is this not a fallacy?

  • 1
    Can you explain what you think "straw-man fallacy" means? It's hard to see how it applies to what Kant means by the CI.
    – virmaior
    Sep 29, 2017 at 1:43
  • Why does something that can be universalized have to be a "simple one-sentence principle"? Why can't it be as complex as one wishes? There seems to be no connection between complexity of ethical maxims and acceptability of the world where everyone follows them. Lying to a murderer, or lying to reduce somebody's suffering, etc., are perfectly universalizable (whether or not they are objectionable on other grounds).
    – Conifold
    Sep 29, 2017 at 18:28
  • Why does something that can be universalized have to be a "simple one-sentence principle"? Why can't it be as complex as one wishes? There seems to be no connection between complexity of ethical maxims and acceptability of the world where everyone follows them. Lying to a murderer, or lying to reduce somebody's suffering, etc., are perfectly universalizable (whether or not they are objectionable on other grounds). Related question What are some examples of categorical imperatives relevant to modern ethics?
    – Conifold
    Feb 12, 2018 at 5:10
  • I feel Kant made a mistake when he decided that lying would always contradict the CI. The CI says nothing about lying. It is such a bad mistake that it is indefensible. It renders the CI useless. But as it stands , as it is written, the CI is easily capable of universalisation.
    – user20253
    Feb 12, 2018 at 13:02

5 Answers 5


In the original question, this was specifically about a "strawman fallacy" which I don't think is the right term. As modified and then answered by Geoffrey Thomas, I think this get's closer to a common objection to Kant that Kant scholars have had to grapple with.

In it's current form, the first paragraph states,

Kant argues that one should only act according to those principles which can be universalised. However human actions are often complex and motivated by multi-layered reasonings. They can rarely be distilled down to a simple one-sentence principle.

The first sentence is just simply what is meant by the universalization formulation of the Categorical Imperative. You can find it in the Groundwork and elsewhere.

The second and third sentences raise a claim that is somewhat debatable, because much hinges on how we describe human actions. Just to illustrate, Utilitarians are not committed to this since they think people act for pleasure (which need not be complex). Aristotelians and virtue ethicists tend to have a complex picture of action.

A second issue is whether we should understand actions in terms of motivations. The other option is to understand them in terms of reasons. The motivation picture is more Humean than Kantian (see Marcia Baron, Kantian Ethics Almost without Apology and Christine Korsgaard).

The relationship between these ideas is complex, but it seems like Kant believes it is possible to have a moral action which finds its origin in the will acting in accordance with reason. (This can be accompanied but not caused by a feeling or subjective motivation for Kant).

If this is what the objection is getting at, then it may be the case that the OP and Kant just have different ideas about what action is.

The second paragraph, however, highlights a real problem in Kant's philosophy. Kant never really gives us a clear explanation of what a "maxim" is or how it relates to one's action and how it becomes universal. Consequently, it is a real question whether we can pick the descriptions of the actions we undertake.

Allen Wood, Henry Allison, and Thomas Hill (inter alia) have all worked on this problem. My memory may be a little fuzzy here but the parts that I found convincing in this account are :

You cannot maxim shop -- it is not a question of picking a maxim that could explain your action and be universalizabe. It's a legitimate question of what you wanted to do when you acted. This seems pretty convincing to me, because maxims are not alibis; they are descriptions of the structure behind your actions.

Actions do admit some multiple descriptions the same action could be described in multiple ways. Kant admits the same in the quodlibetal questions that follow each section of the Metaphysics of Morals: Doctrine of Virtue.

Here, we have Kant saying that banter that includes untruths like "that's a splendid dress" or "of course you look fine" are not lies. Or similarly that willing that you die to protect others is not suicide. Or that killing in war is not murder.

The danger is that once you open the door to these sorts of re-descriptions, it's not clear how you justify some and not others. In my view, Kant does not have great resources for defending the quodlibetal questions. This is a central part of Hegel's critique of Kant (in Natural Law, again in Phenomenology of Spirit, again in Philosophy of Right) -- that "pure reason" alone is not adequate to create descriptions of real human situations and that we need societal knowledge to bridge the gap.

Many contemporary Kantians agree. Some of them argue that Kant was already saying this (Korsgaard, Rawls, Nancy Sherman, Thomas Hill, and Allen Wood -- who interestingly says this in conversion from the Hegelian position). Others admit Kant did not say this but don't care (Habermas).

To recap, there's two issues going on here:

  1. Kant's theory of action vs. a common contemporary one.
  2. The problem of "maxim shopping" with the partial response -- it's the reasons for which we actually undertake our actions plus the problem of description.
  • A recent review is The Problem of Relevant Descriptions and the Scope of Moral Principles by Schumski, who credits Anscombe's Modern Moral Philosophy (1958) with coining the name at the root of modern objections:"rule about universalizable maxims is useless without stipulations as to what shall count as a relevant description of an action with a view to constructing a maxim about it." Schumski recommends dropping that "universal" must hold for all instances of action type in it.
    – Conifold
    Feb 12, 2018 at 5:28
  • It's 'Geoffrey' btw. not 'Gregory'. Excellent answer.
    – Geoffrey Thomas
    Feb 12, 2018 at 6:17
  • @GeoffreyThomas Thanks. I've corrected the misspelling of your name. Considering moving closer to linking this to my actual name but a bit wary to take the plunge.
    – virmaior
    Feb 12, 2018 at 6:49
  • @virmaior. That's fine. You must decide on the use of your name. I rather like 'virmaior'. It is Latin - vir maior - 'a greater man', referring to Kierkegaard ? This perhaps isn't the place to ask but how can I change my profile ? It's probably simple but it defeats me ! Best - GT
    – Geoffrey Thomas
    Feb 12, 2018 at 8:38

You raise an important point. It is reasonable to suppose that any action has a variety of descriptions. If I turn on the light in a room (1) I move my hand, (2) I activate an electrical current, (3) I illuminate the room, (4) I alert a burglar ... All these are perfectly proper descriptions of what I do.

This translates straight across to morality. If I deliberately tell an untruth, (1) I lie, (2) I save someone's hurt feelings, (3) I keep a promise of secrecy, (4) I withhold information about myself that someone has no right to know. All of these can properly describe my action - depending of course on the circumstances.

Kant never explains why just one of these descriptions, 'I lie', should be the dominant description, the one to which I apply the universalisability test.

One might say that Kant commits a fallacy in the very loose sense that he infers from his ethical theory that 'I lie' is the dominant description when this doesn't logically follow from his theory. But I would prefer to say just that he opts for one description of an action and decides, without solid justification, that, no question about it, this is the description to which the universalisability test applies.

It is fair to point out that in the 'Metaphysics of Morals', as distinct from the 'Groundwork', Kant shows himself to be sensitive (at least moderately) to the alternative descriptions under which an action can be morally assessed.

  • 1
    I think that especially when arguing with Davidson, one should note that the notion of primary reason corresponds quite well with Kant's notion of a maxim being the subjective principle of the action, i.e. the reason one subjectively thinks to be the reason (although there may be and are others), and therefore to be seen as the cause of the (physical) action. Davidson actually is quite Kantian (not only in) in these accounts. Additionally, I'd think it's fair to reference his 'Actions, Reasons, and Causes', Journal of Philosophy, 60: 685–700, 1963, esp. when using his example.
    – Philip Klöcking
    Feb 11, 2018 at 13:36
  • @ Philip Klöcking. I agree but isn't there still the fact that an action, hence a maxim, is capable of a variety of descriptions ? What one agent takes to be 'the' reason - the dominant description - may not be the dominant description of the same action for another agent. I took the questioner to be asking how, say, lying could be unqualifiedly ruled out simply because Kant chooses to use one dominant description rather than another.
    – Geoffrey Thomas
    Feb 11, 2018 at 14:13
  • @ Philip Klöcking. I don't think btw that the accounts of different descriptions of the same action are exactly the same in Davidson and Anscombe. But I could not go into this without complicating the answer unduly.
    – Geoffrey Thomas
    Feb 11, 2018 at 14:18
  • Well yeah, that's all fine. But still, there are two objections: First, Kant uses examples, being open to other persons having other maxims and only insisting that persons, in the end, have their reason they think is the main one - just as Davidson (although looking in detail, Kant uses a certain level of abstraction that may be incompatible with Davidsonian primary reasons). Second, while one of the first authors pointing out the complexity of intentionality, Davidson has been widely criticised exactly because he is very Kantian in his notion of primary reasons.
    – Philip Klöcking
    Feb 11, 2018 at 14:22
  • @Philip Klöcking. I don't think I need to rest my claim of different descriptions on Davidson's or Anscombe's accounts. I did not see my citation of Davidson as locking me into a discussion of the philosophical psychology of which in Davidson the idea of different descriptions forms a part. But as ever I appreciate your wide and acute knowledge of Kant - the best on PSE - and do not deny the connexions you drew out from Davidson given the fact that I had cited him. Best - GT
    – Geoffrey Thomas
    Feb 11, 2018 at 15:00

Kant's Categorical Imperative relies on what is logically possible. Even if we take your position that it is only okay to lie to a murderer (and not at any other time), Kant can still construct a logical contradiction. Namely:

Assume everyone lies to murderers. Therefore, everyone would know that anything said to a murderer is false. As a result, the information that one might wish to conceal by lying is laid bare because it is simply the opposite of what was said. Therefore, your statement is no longer a lie since the conveyed meaning of the statement within its linguistic and social context is in fact a truth. This is a contradiction because we assumed that the statement was a lie.

At this point, your more subtle argument comes into play. "There are many ways to lie," you might say, "Therfore, there is no way to discern the truth merely from the knowledge that the statement is untrue." This is analogous to your position that motivations are complex and that context matters. Maybe it can be okay to lie in some situations (or to some questions) but not in others.

The answer to this criticism is that not every situation that is willed universally must create a logical contradiction. As long as any do (such as being asked a strictly yes-or-no question by the murderer), then it is disallowed. Therfore, it is not a straw man fallacy because it is more nuanced and flexible then you are giving it credit for. It was actually designed to accept different situations or different contextual details as parameters.

The upshot is that lying is pretty much never allowed under Kant's arguments, but other things (such as killing another human being) can sometimes slip through depending on the context. Although, off the top of my head, I can't think of a good example where the CI would allow for murder.


The thing is the Cognitive Imperative is a rule used to govern other non-universal rules. For instance one version of Kant's CI states that humans must not be seen as a means to an end, but must instead be considered an end themselves. Essentially it is a rule stating that one cannot use other people merely for personal gain. This rule can be used to govern all other rules without contradiction.

On the other hand if your rule requires adherence to multiple specific behaviors, rather than an abstract goal like in the CI, it can be shown to be false. For instance one may believe it is universally wrong to lie, and universally wrong to kill. However one can then face a situation where if they do not lie, someone will die. In this scenario no course of action would be considered acceptable, and thus the ethical system demanding one never lies and never kills is illogical.

  • 1
    Why am I downvoted?
    – Braydon
    Sep 29, 2017 at 6:00
  • 1
    Probably because there are several inexact statements in there, the example is highly controversial, and it is completely missing any reference to any source for the claims made. These are three possible and valid reasons for downvotes. Additionally, votes are anonymous and nobody has to justify them. I didn't vote down, but can totally understand why one would.
    – Philip Klöcking
    Sep 29, 2017 at 16:06

I do not think Kant would take issue with your position on the basis that you must think lying in general is okay; rather, he would reaffirm the logical integrity of his own rule or measure, preferring logical integrity as a basis of decision making, over and above utilitarian considerations. I do not think he would be committing a logical fallacy, since I do not think he would take issue with your position by means of accusing you personally of lacking personal integrity - simply that he would prefer his own measure because, in contrast to utilitarian considerations, it generates a rule of logical consistency and integrity and upholds a common picture of what every person is owed by every other person (say, truthfulness), ruling out from the start any deviations from that measure

  • Maybe read Kant first? You dont Seem to know his work
    – Saa
    Sep 29, 2017 at 3:06
  • You Also keep using the word utilitarismen as if that is the only alternative here
    – Saa
    Sep 29, 2017 at 3:08
  • Sorry; I acknowledge that utilitarianism was not even a "position" in philosophy when Kant was writing - but I do think perhaps Kant's position on ethics could have been partly generated by a concern about the implications in general of basing decisions on purely external considerations, so instead he went the other way by generating a universalis-able single rule, taken from the point of view of any actor and giving them an epistemology which could be applied in any situation, instead of being dependent on individual circumstances and on the calculation of material benefit to different actors
    – l_ruth_
    Sep 30, 2017 at 23:12

You must log in to answer this question.

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