> A rule by its nature is incapable of capturing
>and accounting for all the morally salient features of a situation.
>So while Kant might be happy to tell us never to lie, even at the
>cost of showing the murderer at our door the way to his victim’s
>room ….

It’s true that Kant did, in a pamphlet, say we shouldn’t lie even to the murderer at the door, but his theory has better resources for dealing with this kind of situation than he applied to it. Several issues need to be disentangled.

The first is that there is good sense in saying that morality involves categorical imperatives – which many Kant interpreters have confused with other issues. What’s sensible in it is the denial that moral requirements can depend on desires (passions, goals, etc.) that people merely happen to have. The reason that I shouldn’t kill you is surely not (just) that I don’t want to. If that were all there is to it, then I could lose the obligation if my desires were to change. If the relevant desires are under my control, I could even bring it about that I have no obligation to spare your life. [Saying that moral requirements are categorical is not equivalent to saying that desires (etc.) are irrelevant to what one ought to do; it is just that they are not all that is relevant.]

Note that the claim that moral considerations or requirements are categorical is a substantive claim about those considerations, though it does not say anything directly about the correct moral content. It is a substantive claim because it can be distinguished from the positions of people like Phillippa Foot at the time of “Morality as a System of Hypothetical Imperatives.� If the “categoricalness� claim is correct, views according to which morality is hypothetical and depends on some desire or goal that we merely happen to have or to have chosen are mistaken. Nonetheless, though the claim for categoricalness is substantive, it does not speak directly to the content of morality. It says instead that whatever the correct content, obligations to pursue, promote, or protect what is specified are not conditional upon happening to have some relevant desire (etc.) that is served thereby. In this sense, there could be, for example, a categorical requirement that people pursue their own interests.

Now, I take the substantive claim that moral requirements must be categorical to be correct, but it is often confused with a formal claim – which can also be found in Kant (and perhaps the confusion, too) to the effect that adequately specified moral principles must be universal in form. The idea can be expressed intuitively this way: It can’t be that two different agents with the same desires, knowledge and so on should be in qualitatively identical circumstances and have differing moral obligations – that one ought to do x while the other ought not. If either has the moral obligation, M, in those circumstances, then they both do. If either does not have the obligation, then neither has it. Somewhat more formally, this can be put by saying that adequately specified moral principles may contain no proper names or unbound agent-variables. (Of course, we can admit that it is enormously difficult to adequately specify a moral principle and that, in practice, we may be stuck with just trying to make sure that our conclusions do not depend upon non-universal considerations.) This, I think, is a matter of the logic of moral terms which should be accepted on any theory. (Dancy seems to disagree, but I have so far been unable to make sense of what he thinks is wrong with it.) Abstractly, the claim would be that the general form of a moral principle is this:

(x) (Fx --> ($y)(Gy & Oxy))

Or: for anyone, x, who has a set of properties, F (including both characteristics of the agent and features of the circumstances in which she acts), there is some act, y, with the set of properties, G, such that she ought to perform y. (Simple manipulations of this format will yield prohibitions [ought-nots] and permissions [not-ought-nots].)  The relevant alternative to universality is particularity – where the alleged principle would include and depend upon the presence of proper names or unbound agent-variables. A principle that satisfies the universality requirement may be as tailored to circumstances as you please – as long as no restrictions are imposed upon what features may be referred to by the predicate, F.  It may even be uniquely applicable to the circumstances of a given agent at thus-and-such a time and place, but it won’t be uniquely applicable to her because she is that particular agent, then and there. However, the universality requirement would be violated by any principle that included proper names or unbound agent-variables – that said, for example, that

Fb --> ($y)(Gy & Oby)

Or: if some particular agent, b, has certain properties, F, then there is something, y, that b ought to do – and the person who affirms this rejects or tries to remain uncommitted on whether, if a different person, c, also had the set of properties, F, c ought to perform y. (Of course, Fb --> ($y)(Gy & Oby) is perfectly acceptable if it is understood as an instantiation of the first quantified formula above.)

It may be noted that satisfying the universality requirement for moral principles is not the same as satisfying the categoricalness requirement because no restriction – except prohibiting covert proper names or unbound variables – is imposed on what may or must be included among the properties, F. So far as satisfying universality goes, it might be that the predicate F includes nothing but desires (etc.) which the agent merely happens to have.

The universality-particularity distinction is in turn often confused with the generality-specificity distinction. A rule or principle is general if it applies to many cases (“never steal� or “never lie�) and is general to the degree that there are few or no qualifying circumstances that limit its application (“never kill except in self-defense�). A principle or rule is specific to the extent that (when fully stated) it applies to relatively few cases. Once one grasps the distinction, it is easy to see that generality-specificity is a different distinction than universality-particularity. In the first place, the universality-particularity distinction divides principles neatly and exhaustively into those that are and those that are not universal, while the generality-specificity distinction is a matter of degree. Second, a highly specific principle may easily still be universal (“anyone must bury her mother under an oak tree unless ....� – with lots and lots of ‘unless’ clauses) while a highly general principle may fail to be (“anyone named ‘Jones’ may be enslaved by anyone else�).

It is, I suspect, a confusion between the universality-particularity distinction and the generality-specificity distincttion – perhaps with an assist from his Pietist upbringing – that led Kant to think that correct moral rules must be highly general, easily stated and exceptionless as stated. But that does not flow from either of the correct claims that morality must be categorical or that it must be universal. Whether the correct principles should admit of short, highly general and exceptionless-as-formulated statement is something that, so far as I can see, depends on further facts.

Having tried to disentangle these matters somewhat, let’s return to the case of the inquiring murderer. Kant actually admits several things that may be relevant to whether he has correctly judged that case:

1. It appears that the maxim to lie when it is necessary to save a life is universalizable. At least, I’ve never seen a convincing counter-argument.

2. Kant allows that, in some cases, though he thinks that it is always wrong to lie, one might permissibly tell a truth (which one knows will mislead) or avoid a direct answer in some way that does not amount to lying. (There’s a story about the fourth-century Christian, Athanasius, that may illustrate this: Athanasius was rowing up-river in a boat when he was passed by Roman soldiers going the other way and seeking to arrest him. They called out, asking if he had seen Athanasius and got the reply, “he is not far away.�)

3. In other places, Kant discusses apparent conflicts between duties – like the duty to tell the truth and the duty to save the life of a murderer’s intended victim (though he doesn’t use that example) – and claims that apparent conflicts should be resolved in favor of the duty with the stronger ground. It appears that might allow a Kantian, if not Kant himself, to conclude that saving a life has a stronger ground than telling the truth and thus could over-ride the apparent duty not to lie.

4. In the Lectures on Ethics, he specifically considers cases in which lying may be justified: “If we were at all times to be punctiliously truthful we might often become victims of the wickedness of others who were ready to abuse our truthfulness .... if I cannot save myself by maintaining silence, then my lie is a weapon of defense.� (228) Since the Lectures is earlier than the pamphlet on not lying from benevolent motives, maybe he changed his view, but it is not clear that anything about the logic of his mature position required the change.

In summary, Kant’s theory, once certain confusions are straightened out, is much better able to handle hard cases like that of the inquiring murderer than Kant himself was (or at least is better able to handle such cases than Kant did).


Rob

---
Rob Bass
rhbass@gmail.com
http://oocities.com/amosapient

A great many people think they are thinking when they are merely rearranging their prejudices.

                            – William James