Many Kantians have reservations about Kant’s rather strident view concerning self-killing. In the Lectures on Ethics in particular, Kant’s rhetoric pulls no moral punches.  Kant calls suicide "revolting", an act wherein we reduce ourselves to mere "carrion" with "no intrinsic worth." We ought to "shrink in horror" at the very thought of suicide, for "nothing more terrible can be imagined" than to treat our own value like that of a beast. 

David Velleman provides one of the few sympathetic defenses of a broadly Kantian view of self-killing. Here I hope everyone might help me put his defense in its most plausible light.

The gist of the Kantian argument against suicide, euthanasia, and self-killing generally is that the rational autonomous self is a source of inherent and unconditioned dignity. Hence, it would be wrong to destroy the physical instrument (the body) through which that autonomous self exercises its rationality.  As Velleman expresses it, Kantian agents have an ‘interest-independent’ value that precludes their self-destruction even if the continued existence of that instrument would result in great suffering or unhappiness for that agent.

It seems obvious that the hardest cases for the Kantian view are those where our continued future existence is bereft of hope (in cases of severe and persistent depression, say) or likely to be suffused with a great deal of pain. Part of Velleman’s defense of a broadly Kantian ethic of self-killing focuses on the latter kind of example. He writes:

"Pain is a bad thing, of course, but I doubt whether it can justify
anything close to euthanasia or suicide unless it is unbearable.  And
then what justifies death is the  unbearableness of the pain rather
than the painfulness. 
What do we mean in calling pain unbearable?  What is it not to bear
pain?  It certainly isn’t a matter of refusing to feel the pain, of
shutting one’s eyes to it, as one might to an unbearable sight, or of
walking away from it, as one might from an unbearable situation.  Not
to bear pain is somehow to fall apart in the face of it, to
disintegrate as a person.  To find pain unbearable is to find it thus
destructive not just of one’s well-being but of oneself.
But then we make a mistake if we describe the patient in unbearable
pain as if he were his old rational self, weighing the harm of pain
against the benefits of existence.  If his pain is truly unbearable,
then he isn’t his rational self any longer: he is falling apart in
pain.  Even if he enjoys some moments of relief and clarity, he is
still falling apart diachronically, a temporally scattered person at
best." ("A right to self-termination?", Ethics 109 (1999), 618)

Velleman doesn’t say much by way of explicating this argument, but the line of thought seems to be this:
a) To undergo severe and unbearable pain (over some specified duration) undermines diachronic personhood.
b) Self-killing is wrong only if the conditions for rational agency are met.
c) Diachronic personhood is among those conditions.
So it is not wrong to engage in self-killing if one is undergoing severe and unbearable pain (over some specified duration).

I have a number of thoughts about this apparently valid argument:
1) Is a) a credible claim?  Velleman says little in its defense, and my familiarity with the personal identity literature is not sufficient for me to say.
2) Does the argument fail to justify prospective self-killing, i.e., killing oneself in advance of the likely pain? The argument seems to depend on pain already being present to undermine diachronic personhood.  It doesn’t claim that a person who, for instance, is diagnosed with a disease that will later prove very painful is not identical to the person who will later suffer the pains. This might seem like a rather cruel conclusion: that one has to suffer great pain in order to justifiably kill oneself, but it would have been wrong to kill oneself in advance of that pain.
3) Suppose that the argument is sound.  Is the following rejoinder available, even to Kantians? "The argument succeeds in showing that self-killing is morally permissible in Kantian terms by showing that the self who kills and the self that is killed are not the same self. But such killings are still wrong on the grounds that one is killing someone else."

Observations and interpretations welcome!

12 Replies to “Velleman on pain, persons, and self-killing

  1. Hi Michael,
    Suppose personal identity requires a continuing subject for consciousness. Suppose unbearable pain occurs only if it is had by a subject of consciousness. I take it then the thought is that when one undergoes unbearable pain, one either goes out of existence and a new subject of consciousness comes into existence, or else one stays in existence, but the unbearable pain somehow dimishes one’s ability to be rational.
    The first horn seem false. It’s rational to fear future unbearable pain, but if one just goes out of existence, it wouldn’t be our own pain that’s feared. That seems wrong. However, if the second horn is correct, and it seems true to me, then (a) is false. At least, we remain the same person through the pain experience even if we are not as rational towards the end of it, which is probably true.
    So, I guess I don’t see why one would appeal to (a) instead of the claim that it’s wrong to kill something only if it’s rational. Then one could argue that during experiences of unbearable pain, one isn’t rational, etc. I don’t think this is a good argument, since it misses the importance of animal pain and cases in which those who undergo unbearable pain are likely to quickly emerge from it and have a good life. In such a situation the killing seems to me to be wrong.

  2. I actually have a lot symphaties with the Kantian argument even though I would probably put in slightly differently. Be that as it may, I have a worry about what the conclusion could be. It is supposed to be that the *act* of suicide is morally permitted when the agent has lost her rational abilities to act due to pain. But, that sounds like a contradiction of terms – like there never could be cases where an agent could justifiedly commit a suicide. If she is able to potentially commit the *act* of suicide, then she must be an agent who can plan actions and excecute them on her own decisions. This sounds like she would be required to have the very capacities that are to be treated as ends in themselves. However, if she has lost the rational capacities and thus is not required to be treated as an end, then it’s not clear how she could commit the *act* of suicide that requires those capacities. I think, in this sense, Kant’s own view that suicides are never morally permitted as acts made sense. This is not to say that it can be amoral when it just happens that a nonrational human causes its own destruction.
    With regards to your questions:
    a) seems plausible to me if by personhood one means the kind of practical rationality as Kantians often say. It seems possible to be in so much pain that one loses any ability to deliberate and act on representations.
    b)I’m not sure it should justify such cases. I’m not sure I have intuitions about this. How much before the pain one could kill oneself? An hour? An week? A year?
    c)I don’t think the rejoinder is available. The someone else one would be killing would not be a member of kingdom of ends and thus there is no direct duty to not to kill them. Of course, there is the indirect argument against killing animals – that it makes one more cruel and likely to kill agents too. But, that doesn’t seem to be worry if one is dead oneself.

  3. Velleman says: “Pain is a bad thing, of course, but I doubt whether it can justify anything close to euthanasia or suicide unless it is unbearable.” I guess this isn’t just supposed to be autobiographical. He means to assert that bearable pain cannot justify suicide. This just seems false to me, if “bearable” means “doesn’t makes me fall apart as a person.” I’ve had pain that, when the doctor asked me “rate the pain on a scale of 1-10,” I said “10.” I was pretty rational at the time (was able to give directions to the hospital, etc). I wasn’t falling apart as a person, if I understand what Velleman means by that. But if I had to face a lifetime of that pain, I think it would be totally rational for me to kill myself. Velleman thinks I’m wrong, I guess, but I’m not seeing the argument.
    As for your first question, the way the terms have been defined by Velleman, (a) is analytic. If we take ‘unbearable’ just to mean ‘really severe,’ then I think (a) is false for the reason just given.

  4. I’m not sure what you mean by “self-killing”. Here’s a natural definition: a self-killing is a killing where the thing that does the killing and the thing that is killed are one and the same. A killing is a self-killing, on this definition, iff a certain relation holds between killer and killed, namely, the relation of identity.
    But at the end of your post you say this:

    The argument succeeds in showing that self-killing is morally permissible in Kantian terms by showing that the self who kills and the self that is killed are not the same self.

    But this seems incoherent given the above definition. If killer and killed are not the same, then it’s not self-killing; it’s other-killing.
    Perhaps you have in mind a different definition: a self-killing is a killing where the thing killed is a self. But on this definition, when, say, Chapman killed Lennon, that was a self-killing, because Lennon was a self (or so I presume; I’m not really sure what it means to say something is a self). So this definition can’t be right.
    This may have some bearing on something else you say:

    b) Self-killing is wrong only if the conditions for rational agency are met.

    This is a bit vague, but I assume you intend it to mean that self-killing is wrong only if the thing killed is a rational agent. But, on the natural definition of self-killing, the thing killed is a rational agent only if the thing that kills is a rational agent (because they are one and the same). And if the killer is not a rational agent, then it’s doubtful that the killing is apt for moral evaluation in the first place.

  5. Smallish point about Kant on suicide. His views are not as absolutist as they might initially seem to be. Thus he held that Cato’s suicide was justified on the grounds that since Cato could no longer be Cato, Cato would not be…. Note that this is consistent with the general grounds of the prohibition.

  6. “The gist of the Kantian argument against suicide, euthanasia, and self-killing generally is that the rational autonomous self is a source of inherent and unconditioned dignity. Hence, it would be wrong to destroy the physical instrument (the body) through which that autonomous self exercises its rationality.”
    Even if the 1st sentence is true I do not think that if has any normative value, it is simply stating a fact; A is the source of x. How does one argue from “A is the source of x” to ‘it is wrong for A to kill herself.’ There needs to be an argument that establishes that the premise ‘killing is wrong,’ is true so that it can be used as a premise in the argument that establishes that the second sentence above is also true, but I do not see how that can be established without begging the question. If this is the case, then it seems that there is no Kantian argument againse suicide, etc.
    Of course, I am not a Kantian so I am probably way off base.

  7. Interesting discussion. Does it actually happen that someone ceases to be a *person* due to pain? I wonder if Velleman’s exception ever obtains in practice.
    Regarding the paradoxical nature of permissibly killing oneself when one is no longer a rational agent, I suspect there are multiple notions of the self in play. I think it’s rational to fear the disintegration of the self that can happen w alzheimer’s or huntington’s, b/c even though you will no longer be *you*, in the thick sense that your identity (as a rational agent) will not be intact, it will still, in a thinner sense, be you that lives through the disintegration of your personality and rational agency. This point I think tracks the distinction between the lit on personal identity (eg, fission cases) and the lit on the “self” (eg, frankfurt).
    In response to Q2, I suspect those of Kantian sensibilities could approve of prospective self-killing on the grounds of preserving one’s dignity. If you have a degenerative disease and know you are about to disintegrate as a person, a condition you may rationally see as incompatible w your diginity, you might (it seems to me) justifiably choose suicide to prevent that–precisely out of respect for yourself as a rational agent who does not wish to be reduced to an infantile state. This seems analogous to the case of Cato.
    Notice, BTW, that Velleman’s argument hinges not on the badness of the suffering, but on the loss of personhood. But extreme pain seems grounds for suicide b/c of what it’s *like* to suffer it, not b/c it undermines your personhood. And the suffering (that it’s from *pain* is purely incidental) seems reason enough for suicide in some cases. See, eg, the horrific case of Vincent Humbert, who “lost the use of his four limbs, his sight, speech and senses of smell and taste” and had his mother kill him:,,1071721,00.html

  8. Great replies to my post. I can’t reply to everything, but I’ll touch on some main themes:
    Mark – Right you are. The way Kant treats casuistical questions concerning suicide in the Metaphysics of Morals suggests that his position is less absolutist than it appears.
    Ben and John – You’re not Kantians, I guess. The claim I made about the rational self doesn’t imply the conclusions about self-killing unless we add (as Kant did) a non-consequentialist picture of moral reasoning. (David Cummiskey has defended the value claims while jettisoning the non-consequentialist part of Kant.) The larger point is that Velleman does intend to defend a Kantian view, so he can’t appeal to the badness of pain and (as Ben put it) the rationality of wanting to avoid such pain as a premise.
    Dan – Again, Kantians can’t appeal to the badness of pain. I’m not suggesting that your view about situations like Humbert’s is wrong, only that Kantians can’t support such a view by appeal to the badness of pain. And as for dignity, I have doubts that this is a properly Kantian conception of dignity. There’s lot of discussion about exactly what Kant meant by dignity (Jyl Gentzler has a very good paper on this topic), but it seems clear that it’s closely linked to our capacity to set ends according to reason. It’s not a physical condition of the body, as is often implied when people speak of ‘death with dignity’. So I guess it’s not obvious that degenerative diseases (not Alzheimer’s say, but cancer) threaten or undermine dignity as Kantians understand it.
    Campbell – I think you’re sort of making the point I was raising with Q3. Rhetorically, I prefer the term ‘self-killing’ to ‘suicide’ for reasons suggested by this article:,0,852262.story?coll=la-news-comment-opinions
    Christian, Campbell, and others – Looking back at the last paragraph I quote from Velleman, it’s actually open to different interpretations: Velleman’s talk of ‘old rational self’ might imply that across periods of severe pain, a person disintegrates into multiple persons. On the other hand,the quote might mean that in such periods, we become non-selves or non-persons. The latter looks more plausible as a way to yield the desired conclusion, since then the entity killed, being a non-self, might permissible be killed.
    Any other ideas regarding 1)? Dave Shoemaker, where are you?

  9. “The larger point is that Velleman does intend to defend a Kantian view, so he can’t appeal to the badness of pain and (as Ben put it) the rationality of wanting to avoid such pain as a premise.”
    I am going to go out on a limb (by showing my ignorance) and ask whether or not a person can set up avoiding pain and seeking pleasure as part of a hypothetical imperative relative to designing a plan for living a good life? It would seem to me that if this is so it would not violate the CI to include in this plan the notion that ending one’s life if certin conditions obtained is one rational response to those conditions. I think I can universalize this as a maxim (all rational people have the right to live and die as they deem reasonable) that all rational people could adopt as long as one is not forced to end one’s life. Would it be choosen under a Rawlsian veil of ignorance?
    I will be honest and admit that I have not studied Kant in over 30 years so I do not remember in detail all his intricate arguments. But, it seems plausible, given the three formulations of the CI that I do remember, that suicide, under well thought out circumstances, does not violate any of them.
    Anyway, this is probably off the point of the post, but I thought it might be pertinent to the discussion.

  10. John,
    Things are complicated here: There’s something of a consensus that (a) Kant fails to show that a suicide maxim violates the Formula of Universal Law, and (b) if any of the formulations of the CI succeed in this regard, it would be the Formula of the End in Itself. Much of the discussion among Kantians has concerned how to hold on to the idea that the rational will must be treated as an end in itself without fully embracing Kant’s strong (but not absolute) prohibition on suicide.
    As to Rawls, etc.: I don’t think there’s any equivalent in Kant to Rawlsian ‘reasonableness.’ For Kant, self-killing is a moral question, to be analyzed in terms of a duty of self-preservation. In contrast, many contemporary Kantians take euthanasia and self-killing to be principally a political question to be settled by the kinds of consdiderations you mention. That’s roughly the tack of the Philosophers’ Brief by Scanlon, Thomson, et al.

  11. Thanks for this interesting discussion.
    Several of these posts raise the problem that an agent who is losing his autonomy, because of “falling apart”, is not in a position to exercise his autonomy by killing himself.
    I think of this problem as the main point of the paper. The title “A Right of Self-Termination?” was meant to raise the question whether a right to suicide can be grounded in a right of self-determination (so that it might be called a right of self-termination).
    The concluding paragraph of the paper is this:

    What I have argued here, however, is that there isn’t a fundamental right to choose between life and death. There may still be a moral justification for death in some cases, but it doesn’t rest on a right of self-determination. And without such a right, the case for legalization must proceed more slowly-far more slowly than The Philosophers would like us to believe.

    This theme was introduced immediately after the passage that Michael quotes in his post:

    I don’t think that we serve the patient well, in these circumstances, by claiming broad rights of self-determination in his name. He may indeed be entitled to help in dying, and he will certainly have to participate in the relevant decisions. But let us keep in mind that these decisions would be premature if the patient were not already in the twilight of his autonomy, where self-determination is more of a shadowy presumption than a clear fact.

    As these passages indicate, my concern with personhood, in this context, is a concern primarily with autonomy, which is the aspect of personhood most relevant to rights of self-determination. And as my reference to the “twilight of autonomy”, I think that autonomy isn’t an all-or-nothing affair.
    Again, I’m grateful for a very interesting discussion.

  12. Postscript:
    Readers might also note this bit of a subsequent paper (“Beyond Price”, Ethics 118 [2008]: 191-212):

    My own view is that hastening death becomes morally appropriate only in the context of deterioration or suffering that compromises autonomy to an extent that can make talk of suicide inappropriate.

    The implication of this view — which I believe but have never stated in print — is that passive euthanasia may be easier to justify than active euthanasia.
    PPS: I am not a regular reader of blogs, so I’d be grateful for a heads-up by email if subsequent comments ask for a response from me.

Comments are closed.