A Bleg: Obligation Dilemmas

As I understand it, an obligation dilemma exists where an agent faces a choice situation in which two (or more) of her available act alternatives are morally obligatory and yet it is impossible for her to perform both of those two act alternatives. Does anyone know of a moral theory or a moral theorist that endorses the existence of obligation dilemmas? Do you think that the divine command theory allows for obligation dilemmas in that it seems possible for God to command an agent to perform both of two mutually exclusive act alternatives? If the divine command theory does allow for the possibility of obligation dilemmas is this a good reason to reject the theory?

57 Replies to “A Bleg: Obligation Dilemmas

  1. Hi Doug –
    Luke Robinson at SMU has done some work on this topic. I’m not sure if it’s published, but he’s got a manuscript floating around out there called “Conflicts of Obligation: A Dispositionalist Account” (I think that’s the title).

  2. I seem to recall an argument attributed to Peter Geach that the only way we could be confident the world would be free of obligation dilemmas is if there were a God ensuring that they don’t arise.

  3. Hi Doug,
    You might want to look at Sayre-MacCord’s “Deontic Logic and the Priority of Moral Theory,” Noûs 20 (1986), pp. 179-197. He argues (I think) that the possibility of moral dilemmas can’t be ruled out a priori (on logical grounds). I’m not sure whether he gives examples of moral theories that generate moral dilemmas. You should also take a look at Michael Slote’s _Beyond Optimizing_ for a utilitarian moral dilemma (I understand that this example was actually Williamson’s). Going back further, Bernard Williams ‘Ethical Consistency’ suggests that there are genuine moral dilemmas, and so does B. van Fraassen, ‘Values and the Heart’s Command’, Journal of Philosophy (’72). Anyway, a few ideas.

  4. Thanks to all of you who have responded so far. I’ll look into these helpful suggestions.
    Mike: Are the moral dilemmas that you’re speaking of obligation dilemmas? I recognize that many people think that prohibition dilemmas (cases where all the act alternatives available to an agent are morally wrong) are possible, but I didn’t know of anyone who had argued for obligation dilemmas. In any case, I don’t see how you could get an obligation dilemma on utilitarianism — the view according to which an act is morally permissible iff no act alternative produces more utility than it does. And the only way that I can see to get a prohibition dilemma on utilitarianism is to hold that the produces-more-utility-than relation is intransitive such that it is possible for a1 to produce more utility than a2 and for a2 to produce more utility than a3 even though a3 produces more utility than a1.

  5. Hi Doug,
    No, I’m talking about obligation dilemmas. Each of these authors suggests that they’re possible. The only person I know that argues that prohibition dilemmas are possible, but not obligation dilemmas, is Peter Vallentyne. But I think the paper in which he does this is not cogent. Anyway, as for utilitarian moral (obligation) dilemmas, I’m pretty sure they’re possible. They will have this form: O(A1 v A2 v . . . v An) & O~A1 & O~A2 . . . O~An. So you ought to perform some available option and you ought not to perform any particular option. The situations that generate such dilemmas are in general a bit artificial, but nonetheless possible. Suppose there is a wall W and a ten foot red line R perpendicular to W. Suppose you will produce more overall happiness the closer you stand on R to W (without touching W); this might involve some story about a device that produces this happiness, or whatnot. If you touch W, you produce no happiness at all, and if you do not stand on R you will produce no happiness at all. It is clear that you ought to stand somewhere or other on the line, since performing A1 v A2 v . . . v An is necessary to maximizing happiness. It is also clear that there is no particular place on the line that you should stand: for any place you choose to stand on R, there is some place closer to W that would produce more overall happiness. But then we have obligations that are impossible to fulfill: viz., O(A1 v A2 v . . . v An) & O~A1 & O~A2 . . . O~An.

  6. John Broome has come up with a very ingenious argument against the view that there can be deontic conflicts of this kind, which he presented a few months ago at a BPhil seminar here in Oxford. As he claims, rationality requires you to intend some act if you believe you ought to do this act. But rationality also requires you not to have contradictory intentions. Broome calls the first of these two requirements ‘Krasia’; let’s call the second ‘Consistent Intentions’. Now assume, for reductio, that it is true both that, as Williams, Slote, and other authors mentioned in this thread believe, you ought to do some act, and that you ought not to do this act. It is then virtually certain that, under at least some circumstances, it would not be irrational for you to believe those truths (there are few if any truths which are not rationally believable under any circumstances). Now suppose you find yourself in one of those circumstances. You would accordingly not be irrational if you believed both that you ought to do some act and that you ought not to do this act. Krasia would then require you to intend to do this act and to intend not to do it. But Consistent Intentions would then imply that you are irrational. This is absurd. So there cannot be any such conflicts.

  7. Mike,
    Interesting case. I’ll need think about it some more. But I’m not quite sure that it’s an instance of what I’m looking for. I’m looking for a case in which an agent faces a choice situation in which two (or more) of her available act alternatives are morally obligatory and yet it is impossible for her to perform both of those two act alternatives. “A1 v A2 v . . . v An” is not an act alternative but a disjunction of act alternatives? Also I would want to insist that there is, in any given choice situation, only a finite number of available act alternatives that the agent can perform. It seems that your example may depend on this being false. Do you know of any examples where there is an obligation to perform two mutually exclusive act alternatives?

  8. Right, the case I presented had infinitely many options. I think I’m not sure what you mean by an ‘act alternative’. My guess is that you mean something like a simple action (though that’s not especially helpful either).

  9. I started thumbing through Patricia Greenspan’s Practical Guilt just last night. Reviews suggest that it can be read as a defense of the possibility of dilemmas, although there seems to be some ambiguity about whether Greenspan is really intending to argue for this or not.

  10. Mike,
    I mean by ‘act alternative’ the genus to which your A1, A2,…An are instances. Alternatively, you can understand an ‘act alternative’ as the sort of things that act-utilitarianism refers to when it is said that according to act-utilitarianism an act is morally permissible iff no act alternative produces more utility than it does.
    So here’s why I think there can’t be any obligation dilemmas on act-utilitarianism. On act-utilitarianism an act is morally obligatory iff it produces more utility than every other available act alternative does. So A1 is morally obligatory iff its outcome, O1, contains more utility than that of every other available act alternative. And A2 is morally obligatory iff its outcome, O2, contains more utility than that of every other available act alternative. And so in order for A1 and A2 to be two morally obligatory act alternatives (an obligation dilemma) O1 would have to contain more utility than O2 while O2 contains more utility than O1. And that’s just not possible. Am I going wrong anywhere?

  11. Mike,
    It seems to me that your example must assume either that matter is infinitely subdivisible or that goodness does not supervene on physical states. All scientists would deny the first assumption, and nearly all philosophers would deny the second.
    Here is a case that presupposes neither of those two problematic views. Suppose it is true that there is an infinite ascent of metaphysically possible worlds, such that for any one you pick, however good, there is always a better one. If God exists and maximizing consequentialism is true, then He may have faced, at the time of Creation, a dilemma of the sort Mike was raising in his example. These examples have been discussed in connection with the problem of evil, but not, as far as I know, in the context of debates in normative ethics. (Incidentally, the assumption that there is no limit to how good a world can be constitutes no adequate solution to the problem, since the positive evil that exists would still be unaccounted for. The problem of evil is not the problem of insufficient good.)

  12. Hi Doug,
    I’m still not sure what you mean by an ‘act alternative’, and I’m honestly not trying to fuss this point. There is an extensive literature on this question. There is a classic issue of Nous (1976) on this point containing some much-neglected material including Lars Bergstrom’s ‘On the Formulation and Application of Utilitarianism’ and J. Howard Sobel’s ‘Utilitarianism and Past and Furture Mistakes’. In the very same issue Krister Segerberg anticipates the infinite utility problems in ‘A Neglected Family of Aggregation Problems in Ethics’. In any case, Bergstrom’s book, The Alternatives and Consequences of Actions started this whole controversy about what counts as an alternative (I have a copy of that book and would be happy to mail you a copy, if you happen to be interested. It’s really difficult to find.). I’m not sure if we could generate a dilemma in the way you suggest. Suppose it is true that, if I were to do A, I would do A, F-ly. It might be true that I must do A to maximize utility. But it might be better to do B than to do A, F-ly. Take this example from Frank Jackson: It is best that I accept the offer to do the book review and review the book. But I know myself. If I were to accept the offer, I would prograstinate and not finish the review on time. It’s better that someone else reviews the book than that I not finish the review on time. In this case, I ought to accept the offer and review it. But it is also true that I ought not to accept the offer. You can see that this need not be a case in which one option is a conjunction of actions and the other not. We could modify the case to read that I ought to accept the offer with gratitude and I ought not to accept the offer. I ought not to accept it because, were I to do so I would not do it with gratitude. That failure of showing gratitude would cause the editor to send the book to a much worse reviewer instead. But then it begins to look like I ought to accept the offer and I ought not to (unless you’re prepared to reject some pretty reasonable closure principles).
    Pablo S., let me think a little more about what you say here.

  13. Mike,
    First, thanks for the kind offer to send me a copy of the book. And, yes, I’m interested and I’ll contact you by email (probably tomorrow) to ask about the details.
    Regarding act alternatives, I agree that it’s a difficult to come upon a precise account. But what I take to be the really essential feature of an act alternative is that it has an outcome. So, actually, I’m fine with saying that accepting the offer is an act alternative and also that accepting the offer and reviewing the book is an act alternative, although the relevant alternatives differ in the two cases. So I accept that an option (or an act alternative) can be made up of a conjunction of actions. Nevertheless, “A1 v A2 v . . . v An” is not an act alternative. It doesn’t have an outcome — that is, there is no one particular way the world will be if the agent performs “A1 v A2 v . . . v An.” So your case above doesn’t seem to me to be an instance of an obligation dilemma, as I’ve defined it.
    In any case, even if we allow that A1, A2,…An could stand for conjunctive acts, it seems that my argument above goes through. Here it is again: Suppose that S can perform A1, that S can perform A2, but that A1 and A2 are alternatives such that S cannot perform both A1 and A2. Now a genuine obligation dilemma would be a case in which S is both morally obligated to perform A1 and morally obligated to perform A2. But, on act-utilitarianism, this is not possible. On act-utilitarianism an act is morally obligatory iff its outcome contains more utility than that of every other available act alternative. So A1 is morally obligatory iff its outcome, O1, contains more utility than that of every other available act alternative. And A2 is morally obligatory iff its outcome, O2, contains more utility than that of every other available act alternative. And so in order for A1 and A2 to be two morally obligatory act alternatives (an obligation dilemma) O1 would have to contain more utility than O2 while O2 contains more utility than O1. And that’s just not possible. Do you see anything wrong with this argument? It seems pretty compelling to me. And if it’s sound, it shows that it’s not logically possible to have obligation dilemmas on act-utilitarianism.
    Regarding the Jackson-type case, I favor the possibilist solution that is defended by Feldman and Zimmerman. I think that you ought to accept that offer and review it.

  14. “On act-utilitarianism an act is morally obligatory iff its outcome contains more utility than that of every other available act alternative.”
    Doug,
    On this definition, if the agent has two options which both promote utility equally, then neither is obligatory. Of the two acts, neither would produce /more/ utility than every other alternative, so neither would be obligatory. That can’t be correct.
    The alternative is to say that acts are morally obligatory iff they promote no less good than any alternative. But on this view, two options which promoted utility equally would each be obligatory – each would produce no less good than any alternative. That is, on this definition, obligation dilemmas are possible for the act-utilitarian.
    Alex

  15. Alex,
    You write: “On this definition, if the agent has two options which both promote utility equally, then neither is obligatory. Of the two acts, neither would produce /more/ utility than every other alternative, so neither would be obligatory. That can’t be correct.”
    Why not?
    If the agent has two options which both promote utility equally, then both are morally optional if the two options are tied for first-place in terms of their utility production. Alternatively, if some other option produces even more utility than these two, then they are both morally wrong.
    You also write: “The alternative is to say that acts are morally obligatory iff they promote no less good than any alternative.”
    Isn’t like saying “The alternative is to say that bachelors are married men”? That’s something you can say but it is, by definition, false.

  16. Doug,
    Yes, you are quite correct on the first point, I hadn’t thought that through properly. I apologise.
    On the second point, I wasn’t aware that the definition of act utilitarianism was this precise. Perhaps that says more about my awareness than anything else. But still, let me try to salvage my basic point: There are views that are very close to act-utilitarianism, and which do allow for obligation dilemmas. (Whether we refer to such views as “act-utilitarianism” or not does not seem to be a particuarly interesting question, at least to me.)

  17. Alex,
    I think that whether act-utilitarianism, as it is typically understood by those who advocate it, can accommodate various sorts of moral dilemmas is an interesting question, because it is important to understand what the implications of this leading moral theory are.
    I agree, though, that whether or not we call the view that “acts are morally obligatory iff they promote no less good than any alternative” ‘act-utilitarianism’ is not an interesting question. And I agree that this view allows for obligation dilemmas. I just don’t find this nearly as interesting since I know of no one who advocates this view. Moreover, I find the view quite implausible precisely because it implies that, in many fairly ordinary instances (those in which A1 and A2 are tied for first-place in terms of their utility production), the agent is both obligated to do A1 and obligated to do A2 even though the agent cannot both do A1 and A2.

  18. On the divine command question. I don’t think it is possible for there to be moral dilemmas on a divine command account (or, better, on the most defensible divine command account). For if A commands B to f, then, unless A is commanding insincerely, A intends B to f. Necessarily, God never has inconsistent intentions, and necessarily, God is never insincere. So, on the divine command account, no incompatible obligations.

  19. This is slightly off topic, but I’d be interested to know if anyone has written about the relationship between obligation dilemmas and I guess what you might call interpersonal obligation dilemmas (i.e., Cases where S ought to A, S’ ought to B, but it cannot be that S ought to A and S’ ought to B). For reasons that aren’t entirely clear to me, people seem far more convinced that there cannot be obligation dilemmas of the usual sort than they are convinced that there could not be what I’m calling interpersonal obligation dilemmas.

  20. Doug,
    I think the basic argument you’ve pushed at a couple of points is sound. But I would like to know what you want to say about the case like God’s creation dilemma, above, where there are an infinite number of alternatives A1, A2, A3… such that A1 < A2 < A3 .... That is, for every act you might do, there is a better one. What's the utilitarian to do in such a circumstance? The problem is particularly pressing if there is a worst alternative but no best one. For you should certainly not pick the worst available alternative A1. But then A2 is the worst of the available alternatives, and you shouldn't pick that. Then you shouldn't pick A3 either ...

  21. Heath, I think that’s what Doug is calling a ‘prohibition dilemma’. Every option is prohibited because it’s worse than some alternative.

  22. Heath,
    Campbell is right; that’s a prohibition dilemma.
    My view is that act-utilitarianism cannot accommodate obligation dilemmas but can accommodate prohibition dilemmas. In instances where there is an infinite number of successively better outcomes available to an agent, act-utilitarianism will hold that there is a prohibition dilemma. Act-utilitarianism may also be able to account for prohibition dilemmas in situations where there is only a finite number of outcomes available to an agent, but only if the better-than relation is intransitive.

  23. Can I pick up where Mike Almeida left off?
    Doug, you seem to have two arguments that Mike’s example doesn’t yield an obligation dilemma.
    First, you say that disjunctions of actions don’t generally count as actions. I must say that I don’t see this at all. Your reason is:

    But what I take to be the really essential feature of an act alternative is that it has an outcome… So I accept that an option (or an act alternative) can be made up of a conjunction of actions. Nevertheless, “A1 v A2 v . . . v An” is not an act alternative. It doesn’t have an outcome — that is, there is no one particular way the world will be if the agent performs “A1 v A2 v . . . v An.”

    I think Mike has a very straightforward answer. Suppose there is a particular outcome of A1, call it O1; and a particular outcome of A2, call it O2; and so on. Then there will be a particular outcome of A1 v A2 v . . . v An, namely, O1 v O2 v … v On.
    Let me make a general point about this. Actions aren’t like sentences, so it’s kind of doubtful that some of them (as opposed to our linguistic descriptions of them) are disjunctive, some not. If (as Mike suggested) there are basic actions, or if you can identify atoms in some other way, then maybe some such distinction can be made out, but I’d like to see it.
    Second, you say:

    In any case, even if we allow that A1, A2,…An could stand for conjunctive acts, it seems that my argument above goes through. Here it is again: Suppose that S can perform A1, that S can perform A2, but that A1 and A2 are alternatives such that S cannot perform both A1 and A2. Now a genuine obligation dilemma would be a case in which S is both morally obligated to perform A1 and morally obligated to perform A2. But, on act-utilitarianism, this is not possible.

    Okay, but there is an infinite version that looks like it should count as an ‘obligation dilemma’ if the two-option version counts, and you haven’t ruled that one out. Right? To be more explicit: in Mike’s example, there is a countable set, Ω, with each member of Ω obligatory but Ω impossible to fully realize.

  24. Jamie,
    Do you take the act of performing (A1 v A2) at t1 to be an act token, or an act type? I was unclear, but I only want to claim that there are no token obligation dilemmas — that is, cases where two mutually exclusive but available act tokens are both morally obligatory. And isn’t the act of performing (A1 v A2) at t1 the sort of thing that can be instantiated either by performing A1 at t1 or by performing A2 at t2?
    On the second point, could you explain a bit more concretely what the infinite version is. A concrete example would help me immensely. In any case, let me just make clear that it can’t be a case where every available act token is morally impermissible. If it is, it’s what I would call a prohibition dilemma, for an act token is morally obligatory just in case it is permissible and no alternative to it is permissible. I follow Vallentyne (“Two Types of Moral Dilemmas”) here.

  25. Doug,

    And isn’t the act of performing (A1 v A2) at t1 the sort of thing that can be instantiated either by performing A1 at t1 or by performing A2 at t2?

    I’m not clear on whether we have particular acts in mind, A1, A2, or whether this is supposed to be a general schema. What if A1 and A2 are tokens performed at the same time? But I guess I’m not following properly.
    I had in mind Mike’s original example. The acts are these, for each positive natural n: you stand closer than 1/n feet from W on R without touching W. There are countably many of these. Each is obligatory. It is impossible to perform all of them. Doesn’t that example satisfy the criteria?

  26. Jamie,
    I suspect that I’m the one who is not following properly. But let me try to at least make my position more clear. By ‘act alternatives’, I mean ‘mutually exclusive act tokens’. So if A1 and A2 are act tokens (as I was taking them to be in Mike’s example), then (A1 v A2) is not an act token. Also, if A1 and A2 are act alternatives, then A1 and A2 cannot be both performed, let alone at the same time.
    In Mike’s example, I don’t think that each of these acts are morally obligatory, for each of these acts are morally impermissible given the existence of a better alternative in each case. An act token is morally obligatory only if it is morally permissible, and none of these acts are morally permissible given the availability of a better alternative. So since this is a case where all the act alternatives are morally impermissible, it is what I call a prohibition dilemma.

  27. I see. I didn’t understand that the acts have to be exclusive to be alternatives.
    But the Wall example still seems to be in the spirit of an obligation dilemma. The acts in question are not pairwise exclusive, but they are jointly impossible. (Any finite subset could be performed all together, but the complete set cannot be.)
    The acts, stand exactly x feet from the wall for distinct x, are exclusive and all impermissible, I agree. But look at the ones I suggested (stand closer than 1/n feet from W on R without touching W). None of them is impermissible.

  28. Jamie,
    Isn’t the act of standing closer than, say, 1/2 foot from W on R without touching W an act-type? I can instantiate this act-type by performing the act-token in which, at t1, I stand exactly 1/4 foot from W on R without touching W and also by performing the act-token in which, at t1, I stand exactly 1/8 foot from W on R without touching W.
    Your example may be in the spirit of a token obligation dilemma, but it seems to be a type obligation dilemma, not a token obligation dilemma. Again, I want to claim only that there are no token obligation dilemmas — that is, no cases where two mutually exclusive but available act tokens are both morally obligatory.

  29. Your example may be in the spirit of a token obligation dilemma, but it seems to be a type obligation dilemma, not a token obligation dilemma. Again, I want to claim only that there are no token obligation dilemmas — that is, no cases where two mutually exclusive but available act tokens are both morally obligatory.

    Okay, I do see how this rules out the Wall example.
    How do we individuate act tokens? I suppose you mean to individuate them very finely. But then it’s unlikely that there are any obligatory acts (since for any act there will be some trivially distinct one that would have been permissible). If that’s right, then it is a rather trivial fact, too, that there are no obligation dilemmas! (If there are no obligatory acts at all, then there aren’t any situations in which there are two obligatory acts.)

  30. Jamie,
    That’s a good question. I don’t know how to individuate act-tokens. My hope is that those smarter than me working in action theory will come up with a way of individuating act-tokens that vindicates my claim that “standing closer than, say, 1/2 foot from W on R without touching W” is not fine-grained enough to count as an act-token without relying on an account of individuating act-tokens that insists on descriptions that are so fine-grained as to rule out the possibility of obligatory acts altogether.
    Not very satisfying — I know. But, unfortunately, that’s the best I got.

  31. Let me suggest an argument for the view that Mike’s example is both a ‘prohibition dilemma’ and an ‘obligation dilemma’.
    For simplicity, let one metre be the maximum distance you can stand from the wall. For any non-empty subset S of the real interval [0,1], let there be an act A(S), where, for example, A({.5}) is the act of standing exactly half a metre away from the wall, A({x : x < .5}) is the act of standing closer than half a metre, A({.25,.75}) is the act of standing either one quarter or three quarters of a metre away, and so on. Say that one act A(S) is part of another A(T) iff S is a subset of T. And say that A(S) is the complement of A(T) iff S = [0,1] - T. Then the following seem plausible principles. I. If every part of A(S) is prohibited, then A(S) is prohibited. II. If the complement of A(S) is prohibited, then A(S) is obligatory. Now, as already established, every atomic act, i.e. every A(S) with S a singleton, is prohibited, because it has worse consequences than another atomic act. But then it follows from I and II that every act is both prohibited and obligatory. So we have a 'prohibition dilemma' and a 'obligation dilemma' rolled into one. No matter what you do, you'll perform a prohibited act and you'll fail to perform an obligatory act. What if we restrict the set of alternatives to atomic acts only? Doug thinks this will eliminate the obligation dilemma, leaving only the prohibition dilemma. But I'm not convinced. Suppose we define consequentialism as follows: first, an act is prohibited iff its outcome is worse than some alternative, and, second, an act is obligatory iff every other act prohibited. In normal cases, with a finite set of acts, this will behave just as one would expect. But in our example it implies that every act is both prohibited and obligatory. This tends to suggest, I think, that there's no fundamental difference between a prohibition dilemma and an obligation dilemma.

  32. Hi Campbell,
    You write: “If the complement of A(S) is prohibited, then A(S) is obligatory.”
    If you hold that a sufficient condition for an act’s being obligatory is that all of its alternatives are impermissible, then it will, of course, be trivially true that all prohibition dilemmas are obligation dilemmas. I, however, don’t think an act is obligatory so long as all of its alternatives are impermissible. As I noted above, my view is that an act is obligatory if and only if it is permissible and all of its alternatives are impermissible. See Vallentyne’s “Two Types of Moral Dilemmas” for a defense of this conception of an act’s being obligatory. Now if an obligatory act is necessarily a permissible act and a prohibition dilemma is necessarily a choice situation in which all of the available act alternatives are impermissible, then no prohibition dilemma is an obligation dilemma. Since Mike’s example is a prohibition dilemma, it can’t be an obligation dilemma — at least, not if we conceive of an obligatory act in the way I’ve been suggesting.

  33. Doug,
    Wait, something seems to have gone wrong. Here are two definitions of yours (my numerals, your words).

    (1) [M]y view is that an act is obligatory if and only if it is permissible and all of its alternatives are impermissible.

    (2) [A]n obligation dilemma exists where an agent faces a choice situation in which two (or more) of her available act alternatives are morally obligatory

    Suppose there is an obligation dilemma. Then (by (2)) somebody faces a choice between two obligatory acts; call them A and B. But B is an alternative of an obligatory act, and therefore (by (1)) impermissible, and therefore (by (1)) not obligatory.
    So according to your definitions, there are certainly no obligation dilemmas.
    Did you mean for there to be such an easy route to the conclusion you favor?

  34. Jamie,
    Yes. That’s right. Obligation dilemmas are conceptually impossible. Again, I’m just following Vallentyne here. Of course, not all conceptual truths are obvious. And this I think is one of those instances, because it’s not obvious (or at least it hasn’t been obvious to many philosophers) that an obligatory act is necessarily a permissible one. Many philosophers, as all the citations above attest to, have endorsed obligation dilemmas despite the fact that they are conceptually impossible.

  35. The reason that I’m interested in this is because I’m interested in whether the fact that consequentialism cannot accommodate obligation dilemmas is a problem. I would argue that it is not a problem, because obligation dilemmas (although endorsed by many theorists and implied by certain moral theories) are conceptually impossible. This leaves us, then, with prohibition dilemmas. And I think that the consequentialism can accommodate prohibition dilemmas in cases in which there are an infinite number of successively better outcomes that the agent can bring about as well as in other finite-option cases by either holding that the better-than relation is intransitive (as Stu Rachels does) or by holding that whether O1 is better or worse than O2 can depend on whether the agent performs A1 or A2, as our own Campbell Brown has argued.

  36. Fair enough, if “obligatory” is defined as uniquely permissible, then obligation dilemmas are impossible.
    Still, I’m reluctant to accept this definition. I find it very natural and intuitive to hear “You are obligated to do x” as equivalent to “You are not permitted not to do X”. Indeed, it’s quite common in philosophy to treat “obligatory” and “permissible” as duals (like “necessary” and “possible”).
    On the other hand, I also find it very plausible that “obligatory” implies “permissible”. Perhaps, then, the thing to say is this. The following is an inconsistent triad:

    1. An act is obligatory if every other act is prohibited.
    2. An act is obligatory only if it is not prohibited.
    3. Every act is prohibited

    In cases with finite acts, I’m happy to reject 3. But in infinite cases like Mike’s, I’m not sure which of the three to reject. That’s not surprising, however, since Mike’s example does seem rather like a paradox.
    (Incidentally, it turns out now that much of the earlier discussion, about the legitimacy of “disjunctive acts” and so on, was beside the point. If “obligatory” means uniquely permissible, then obligation dilemmas are impossible regardless of what counts as an “act alternative”.)

  37. Doug,
    On your definition of act-utilitarianism,

    an act is morally obligatory iff it produces more utility than every other available act alternative does.

    As you note, on this definition there cannot be obligation dilemmas. Since it is logically impossible for there to be two acts each of which produces more utility than the other, there can be at most only one obligatory act. But since obligation dilemmas can only arise when at least two acts are obligatory, such dilemmas cannot arise on act-utilitarianism.
    Interestingly, this argument need not assume that, as you assume, an act is impermissible if some alternative act is obligatory. All it needs to assume is that the relation produces more utility than is irreflexive, which it clearly is.
    I don’t think your definition should be accepted, though. To the extent that some people find utilitarianism appealing, the appeal of this theory lies, I believe, in the way it captures the intuition that one is never permitted to do what wouldn’t make things go best. Yet this intuition is not fully captured by your definition. When two or more acts are tied for maximal utility production, the theory implies that, of all the available acts, no one is impermissible. Yet if some of them are worse than others, then it follows that it is not wrong for you to do what wouldn’t make things go best. So your definition is inadequate as a formulation of utilitarianism, and a fortiori act-utilitarianism.
    I propose instead that

    On act-utilitarianism an act is morally permissible iff it produces as much utility as any other available act alternative does.

    (Perhaps this is the formulation Alex was trying to get at when he attempted to revise yours.)
    Here it follows that there cannot be obligation dilemmas when there is one act which produces more utility than all the alternatives. For the theory in these cases implies that all these other suboptimal acts are not permissible. From this it follows that they are not obligatory. So there is at most one obligatory act, which for the reasons noted above excludes the possibility of obligation dilemmas.
    Yet when two or more acts produce more utility than all its alternatives, the theory does not rule out such dilemmas. Within the class of alternative acts that are tied for maximal utility production, it is consistent with the theory that two are each obligatory. This becomes impossible only if we assume, as you do, that if some act is obligatory the all its alternatives are impermissible. But on that assumption, obligation dilemmas would be ruled out by all consistent moral theories.
    The upshot, I think, is that there is no interesting sense in which act-utilitarianism excludes obligation dilemmas. Either you assume that such dilemmas are conceptually impossible, in which case it would follow only trivially that those dilemmas are excluded by act-utilitarianism; or you provide a definition which, even without making any further assumptions, rules out such dilemmas, but at the cost of not capturing the essence of utilitarianism; or you provide an adequate definition of the theory which, however, allows for the possibility that there are obligation dilemmas.

  38. Correction: ‘irreflexive’ should have read ‘asymmetric’ in my comment above. It is of course true that the relation is irreflexive, but what is needed for the argument is that it be asymmetric. This point has no implications for what I say in the rest of the comment.

  39. Pablo,
    I assume that you meant to write “at least as much utility” where you’ve written “as much utility” in your proposed alternative definition.
    If so, our two definitions will yield different verdicts only if the ranking of outcomes in terms of their utilities is incomplete — that is, only if there are some incommensurable outcomes. Given incompleteness, your definition says that an act, A1, is impermissible if its outcome, O1, is incommensurable with what would otherwise be the outcome with the most utility, say, O2. But I find it odd to say that A1 is impermissible even though it is not the case that its outcome, O1, is less good than O2. As Campbell Brown has written: the “core idea of consequentialism is that rightness and wrongness are determined by goodness. Hence, if goodness is indeterminate, then rightness should be too. If goodness were indeterminate, how could it determine anything?” So I think that my definition does a better job of handing cases where completeness doesn’t hold. And where completeness does hold, there is no substantive difference between are two definitions.

  40. Doug,
    Yes, I meant “at least as much utility” in my proposed definition.
    I must confess I fail to see why, as you write, “our two definitions will yield different verdicts only if the ranking of outcomes in terms of their utilities is incomplete”. Suppose A1 produces exactly as much utility as A2 does, and that each produces definitely more utility than every other available alternative acts A3, A4,…, An do. On your definition of act-utilitarianism,

    an act is morally obligatory iff it produces more utility than every other available act alternative does.

    Given the assumptions above, there is no act that can be truly claimed to produce more utility than every other available act alternative. A1 isn’t, because A1 produces no more utility than A2 does. A2 isn’t, because A1 produces no more utility than A2 does. And none of A3, A4,…,An are, since they produce no more utility than either A1 or A2 do. So, given your definition, there is no act which is morally obligatory. And since your definition only makes a claim about obligation, it leaves no room for drawing any other morally relevant difference between these acts. It follows that A1, A2, A3,…, An are all morally on a par.

  41. Sorry again for having to slightly amend what I said earlier. When I wrote

    A2 isn’t, because A1 produces no more utility than A2 does.

    I obviously intended to write

    A2 isn’t, because A2 produces no more utility than A1 does.

    The correction should be obvious, but just in case…

  42. Pablo,
    Okay, I see the problem. I should of said earlier that I never claimed that the following was a definition:
    (1) “an act is morally obligatory iff it produces more utility than every other available act alternative does”
    I think that if you look above, you’ll find that I said only that (1) is something that’s true on the definition of act-utilitarianism. I never claimed that (1) was itself a definition. Here’s the definition that I accept:
    (2) x is act-utilitarianism =def. x holds that the deontic statuses of actions are completely and solely determined by the utilities of their outcomes, such that an act is morally permissible if and only if no available act alternative produces more utility than it does.
    (2) differs from your definition only given incompleteness. In any case, my argument doesn’t depend on (1) being a definition. It depends only on (1) being true. Do you deny that (1) is true? I think that I’m right in think that (1) follows from (2). Do you disagree?

  43. Doug,
    Let’s assume completeness. As you write, in these cases my definition and your (2) have the same implications. Now there are two relevant classes of cases. First, there are those in which one and only one act produces more utility than all its alternatives. Here our definitions imply that this act is obligatory, since that act is permissible and all the others are impermissible. So far, so good. But, second, there are those cases in which two or more acts produce as much utility as all the others. My example of A1 and A2 was one of these cases. For (1) to follow, here you need (2) to imply that neither A1 nor A2 are obligatory. But is this the case?
    Well, you assume that if one act is obligatory then all its alternatives are impermissible. Given this assumption, it is impossible that either A1 or A2 are obligatory, for reasons discussed previously. So (1) follows from (2) plus your assumption.
    If this is the argument, note however that (1) could not then explain why moral dilemmas are ruled out by act-utilitarianism, as you claimed earlier in this thread. On the contrary, (1) is a consequence of (2) conjoined with your assumption, and from this assumption alone it follows that obligation dilemmas are ruled out by any consistent moral theory. It is your assumption that is doing all the explanatory work here.
    Moreover, I don’t see any reason to accept your assumption. It’s not at all clear to me that it is part of the concepts of obligation or permissibility that, if one act is obligatory, then all its alternatives are impermissible. I don’t think obligation dilemmas can be ruled out by point of logic. For instance, I think it makes sense to worry, like Sidgwick did, that we might have most reason both for doing some act and for doing some incompatible alternative, when only one maximizes my own wellbeing while the other maximizes the aggregate wellbeing of everyone. If your assumption is true, then these cases should be described as cases in which it is permissible for you to do what best promotes either your own good or the general good, impermissible for you to do anything else, but not obligatory for you to do both. But I think there should be at least logical space to state the possibility that we might in fact be in this terrible situation.

  44. Okay, I wish I could blame the fact that I’m not a native English speaker, but it’s become quite obvious that it’s just carelessness on my part. The second of the cases I distinguish in the first paragraph above are those in which two or more acts produce more utility than all the others.

  45. I guess you can blame my English here. Should I have written that two or more acts produce each more utility than all the others? I’m just referring here to cases like my example above, where A1 and A2 are tied for maximal utility production and A3, A4, … , An each produce less utility than either A1 or A2 do. I hope this is clear.

  46. Pablo,
    So you mean that they’re tied for first-place in terms of utility production. It can’t be that each of A1 and A2 produce more utility than all the others. A1 is other than A2, and A2 is other than A1. So your claim would amount to the claim that A1 produces more utility than A2 even though A2 produces more utility than A1.
    In any case, I don’t know whether I ever claimed that “(1) explains why moral dilemmas are ruled out by act-utilitarianism.” I claimed only that, from (1) and the other plausible premises that I stated in my argument, it follows that act-utilitarianism cannot account for obligation dilemmas.
    But I gather you reject (1). Could you then specify some alternative. That is, please fill in the blank:
    According to act-utilitarianism, an act is obligatory if and only if _______.
    Also, what’s your view on how ‘obligatory’ and ‘permissible’ are to be inter-defined. My view is that an act is obligatory iff it is the only permissible alternative available to the agent. But you reject this. So I don’t know what your view is. You told me what your view about permissibility is, but not what your view about obligatory is. Also, since you reject the idea that, if one act is obligatory, then all its alternatives are impermissible, could you give some evidence that this idea is false. Do you find that competent speakers of English sometimes say things like: “You’re obligated to do x, but it is permissible for you to do y instead”? That’s not something I hear people say. I think that my (1) is supported by common usage.

  47. So you mean that they’re tied for first-place in terms of utility production. It can’t be that each of A1 and A2 produce more utility than all all the others.
    I’m glad you got what I meant, but I’m sorry that you still find my revision unsatisfactory. To me, the scope of the term ‘all others’ in the expression ‘two or more acts produce each more utility than all the others’ is contextually restricted to include all acts other than the two or more mentioned earlier in the sentence. But you are the native English speaker here, so if you think this is not true, I defer to your judgment.
    In any case, I don’t know whether I ever claimed that “(1) explains why moral dilemmas are ruled out by act-utilitarianism.” I claimed only that, from (1) and the other plausible premises that I stated in my argument, it follows that act-utilitarianism cannot account for obligation dilemmas.
    You wrote:

    So here’s why I think there can’t be any obligation dilemmas on act-utilitarianism. On act-utilitarianism an act is morally obligatory iff it produces more utility than every other available act alternative does. So A1 is morally obligatory iff its outcome, O1, contains more utility than that of every other available act alternative. And A2 is morally obligatory iff its outcome, O2, contains more utility than that of every other available act alternative. And so in order for A1 and A2 to be two morally obligatory act alternatives (an obligation dilemma) O1 would have to contain more utility than O2 while O2 contains more utility than O1. And that’s just not possible.

    It is clear, I believe, that this argument is intended to explain why there cannot be obligation dilemmas on act-utilitarianism, and that the first biconditional in the quote—which we are calling (1)—plays an essential role in that argument. Now the issue here is how the dialectic of the argument interacts with your later claim that (1) is not a definition of act-utilitarianism, but follows from that definition–which we are calling (2). My point was that (1) only follows from (2) if you make an assumption about the concepts of obligation and impermissibility which, by itself, implies that obligation dilemmas are ruled out by all consistent moral theories. There’s something problematic, I think, about an argument which intends to explain why a conclusion holds about a particular moral theory by essentially relying on an assumption which is itself sufficient to explain why that same conclusion holds about all moral theories.
    But I gather you reject (1). Could you then specify some alternative. That is, please fill in the blank:
    According to act-utilitarianism, an act is obligatory if and only if _______.

    I reject (1) because I don’t think it follows from the correct definition of act-utilitarianism. What I can give you is not an alternative to (1), but what I take to be the correct definition of that theory, and then translate all its implications into the language of obligation. The definition is the one I gave before:

    According to act-utilitarianism an act is morally permissible if and only if it produces at least as much utility as any other available act alternative does.

    As I stressed before, there are two relevant cases to consider. Consider first cases when there is only one act which produces the most utility. In such cases, the theory declares all acts minus one impermissible, and this one act permissible. It follows that it is obligatory not to perform any of those acts, and obligatory to perform this one act. Consider next cases where two or more acts are tied for first-place in terms of their utility production. Here all the suboptimal acts are impermissible, and so it is obligatory not to perform any of them. And, since the optimal acts are the only ones permissible, it is obligatory that some of them are performed. (This follows from the principle I state below, in boldface. ‘Some’ here means ‘at least one, though not one in particular’.)
    Also, what’s your view on how ‘obligatory’ and ‘permissible’ are to be inter-defined. My view is that an act is obligatory iff it is the only permissible alternative available to the agent. But you reject this.
    Yes, I only endorse the conditional form of the statement above. An act is obligatory if it is the only permissible alternative available to the agent. But an act need not be the only permissible alternative available to the agent if it is obligatory. (In fact, the principle I endorse, from which the above conditional statement follows, is more general: It is obligatory for an agent to perform at least one of the acts in a set if all the alternative acts outside that set are impermissible.)
    Also, since you reject the idea that, if one act is obligatory, then all its alternatives are impermissible, could you give some evidence that this idea is false. Do you find that competent speakers of English sometimes say things like: “You’re obligated to do x, but it is permissible for you to do y instead”? That’s not something I hear people say. I think that my (1) is supported by common usage
    No, this is not what I hear people say. But the explanation may be that, as a matter of fact rather than logic, there are no obligation dilemmas. We don’t hear people say such things because people speak truly.
    However, I should make it clear that I don’t have any settled views on the issue. I merely said that I don’t see positive reasons to believe that the claim “all the alternatives to a permissible act are impermissible” is conceptually true, and accordingly asked you to indicate why you thought it was.

  48. Pablo,
    You quote me as saying: “all the alternatives to a permissible act are impermissible.” Where did I say this? If I did say this, it was a mistake. But I don’t think that I said this. Please indicate the date and time of the specific comment that you are supposedly quoting from.
    You write: “Consider next cases where two or more acts are tied for first-place in terms of their utility production. Here all the suboptimal acts are impermissible, and so it is obligatory not to perform any of them. And, since the optimal acts are the only ones permissible, it is obligatory that some of them are performed.” In your example A1 and A2 are tied for first place and all the acts other than A1 and A2 are suboptimal. I agree that in such a case it is obligatory that you perform (A1 v A2). But this isn’t an obligation dilemma. Was it meant to be? I’m confused as to what your point is here.
    Lastly, I don’t know why you are so insistent that I be interpreted as explaining why act-utilitarianism in particular can’t account for obligation dilemmas. My argument was only meant to establish that it is true that act-utilitarianism cannot account for obligation dilemmas, not that only act-utilitarianism is unable to account for obligation dilemmas. I would be just as happy showing this by arguing:
    P1) Obligation dilemmas are conceptually impossible.
    P2) No coherent moral theory can accommodate that which is conceptually impossible.
    P3) Act-utilitarianism is a coherent moral theory.
    C) Therefore, act-utilitarianism cannot accommodate obligation dilemmas.

  49. Doug,
    The statement which I saw no reason to believe was conceptually necessary was, of course, that “all the alternatives to an obligatory act are impermissible”, which was the assumption that has been the focus of a large part of our exchange. I don’t know why I substituted ‘permissible’ for ‘obligatory’ in the last paragraph of my previous comment. No one to my knowledge has ever contended that “all the alternatives to a permissible act are impermissible” holds by conceptual necessity. So I don’t know why I wrote something that didn’t express what I intended to say. I’m sorry for creating unnecessary confusion.
    You ask: “In your example A1 and A2 are tied for first place and all the acts other than A1 and A2 are suboptimal. I agree that in such a case it is obligatory that you perform (A1 v A2). But this isn’t an obligation dilemma. Was it meant to be? I’m confused as to what your point is here.”
    You are right that this isn’t an obligation dilemma, but it wasn’t meant to be one. The point, stated in the preceding paragraph, was to translate all the implications of my definition into the language of obligation, since you insisted that I use this modal operator in characterizing act-utilitarianism.
    You also say: “Lastly, I don’t know why you are so insistent that I be interpreted as explaining why act-utilitarianism in particular can’t account for obligation dilemmas. My argument was only meant to establish that it is true that act-utilitarianism cannot account for obligation dilemmas, not that only act-utilitarianism is unable to account for obligation dilemmas.”
    I don’t think I insisted that, on your view, only act-utilitarianism is unable to account for obligation dilemmas. Rather, I insisted that your “argument [wa]s intended to explain why there cannot be obligation dilemmas on act-utilitarianism”. I insist in believing this because you prefaced the argument with the sentence: “here’s why I think there can’t be any obligation dilemmas on act-utilitarianism.” In any case, what matters philosophically here is not who intended what, but whether there is any philosophically interesting connection between the fact that act-utilitarianism rules out obligation dilemmas and the claim—which we called (1)—that on act-utilitarianism an act is morally obligatory iff it produces more utility than every other available act alternative does. I contend that there is no such philosophically interesting connection.

  50. Pablo,
    You earlier wrote: “There’s something problematic, I think, about an argument which intends to explain why a conclusion holds about a particular moral theory by essentially relying on an assumption which is itself sufficient to explain why that same conclusion holds about all moral theories.” Weren’t you implying here that I was trying to explain why act-utilitarianism in particular is unable to account for obligation dilemmas? If you agree that my “argument [wa]s intended to explain [only] why there cannot be obligation dilemmas on act-utilitarianism” and not why act-utilitarianism in particular is unable to account for obligation dilemmas, then there’s nothing problematic about it, right?
    Okay, you want to know why I accept that “all the alternatives to an obligatory act are impermissible.” Well, it seems to me that an act is obligatory only if it is wrong to omit it. And it’s wrong to omit an act only if none of its alternatives are permissible. And none of an act’s alternatives are permissible only if all of its alternatives are impermissible. So an act is obligatory only if all of its alternatives are impermissible. Which premise do you reject?

  51. Doug,
    You wrote

    Weren’t you implying here that I was trying to explain why act-utilitarianism in particular is unable to account for obligation dilemmas?”

    No, my point was that your argument had the following problematic structure: it attempted to explain why some members of a class had some property by relying on a premise that could only be established by making an assumption from which alone it followed that all members of that class had that property. (‘Some’ is here used in the logical sense of ‘one or more’.) It is irrelevant for this objection whether you intended the argument to explain why there cannot be obligation dilemmas on act-utilitarianism, or whether instead your intention was to explain why act-utilitarianism in particular rules out such dilemmas.
    You also wrote:

    Okay, you want to know why I accept that “all the alternatives to an obligatory act are impermissible.” Well, it seems to me that an act is obligatory only if it is wrong to omit it. And it’s wrong to omit an act only if none of its alternatives are permissible. And none of an act’s alternatives are permissible only if all of its alternatives are impermissible. So an act is obligatory only if all of its alternatives are impermissible. Which premise do you reject?

    I don’t reject any of the premises. I asked for a reason to believe that all the alternatives to an obligatory act are impermissible, and you have given me an excellent one. As a moral agent, I’m relieved: you have convinced me that, however demanding morality might be, it will never require me to do both some act and one of its alternatives. I’m deeply grateful.
    A question: do you take obligation dilemmas to be ruled out by the logic of the deontic operators alone, or do you think that specifically moral concepts play a role in the argument? Would you, in other words, accept a parallel argument where ‘impermissible’ is substituted for ‘wrong’ in the first premise above?
    I should also correct something I said earlier. I endorsed the principle that it is obligatory for an agent to perform at least one of the acts in a set if all the alternative acts outside that set are impermissible. As stated, however, this principle is too strong, and rules out prohibition dilemmas. The principle implies that if all acts minus one are impermissible, then this one act is obligatory. Since all obligatory acts are permissible, it follows that, if all acts minus one are impermissible, then this act is permissible. Suppose next that a prohibition dilemma obtains, i.e., that all acts available to the agent at a given time are impermissible. Consider then an arbitrary subset of all these acts minus one. Given my principle, this act must be permissible, since all its alternatives are impermissible. But then it follows from the assumption both that some act is permissible and that this same act is not permissible. This is absurd. So, if my principle is true, there are no prohibition dilemmas.
    I think the principle should be revised to read: “when some of the available acts are impermissible, it is obligatory for the agent to perform one of the permissible alternatives.” This formulation doesn’t assume that there are permissible alternatives, and so it doesn’t rule out prohibition dilemmas.

  52. Pablo,
    You write: “No, my point was that your argument had the following problematic structure: it attempted to explain why some members of a class had some property by relying on a premise that could only be established by making an assumption from which alone it followed that all members of that class had that property.”
    Again, I wasn’t offering an explanation as to why act-utilitarianism has the property of not being able to accommodate obligation dilemmas. For instance, I wasn’t arguing that there was something specific to act-utilitarianism (as opposed to something about consistent moral theories in general) that accounts for its inability to accommodate obligation dilemmas. I was merely arguing that act-utilitarianism cannot accommodate obligation dilemmas. What’s the problem with my argument? Is it invalid? Are any of the premises false? Does it beg the question?
    Do you think that the following argument is problematic, and, if so, why?
    1. All consistent moral theories are unable to accommodate obligation dilemmas.
    2. Act-utilitarianism is a consistent moral theory.
    3. Therefore, act-utilitarianism is unable to accommodate obligation dilemmas.
    You also write: “A question: do you take obligation dilemmas to be ruled out by the logic of the deontic operators alone, or do you think that specifically moral concepts play a role in the argument?”
    The former.

  53. Doug,
    Your argument was meant to explain why there cannot be obligation dilemmas on act-utilitarianism. I regard the argument as a failure in achieving this intended explanatory purpose. It fails in this repsect, I believe, because the correct explanation of the proposition to be explained is given by general facts about the consistency of moral theories, and the conceptual truth that, if an act is morally obligatory, then all its alternatives are morally impermissible. Instead of citing those facts, your argument cites facts specific to act-utilitarianism. This is why I believe it is explanatorily defective.
    The argument your provide above, by contrast, cites the relevant explanatory facts, or at least most of them. Although both arguments are sound, only this second argument has the relevant epistemic virtues. Accordingly, I see nothing problematic about it.

  54. I wonder if Kierkegaard’s “Fear and Trembling” discussion of Abraham and Isaac has any bearing? Kierkegaard presents Abraham’s situation as a dilemma between two distinct kinds of obligation–an ethical obligation to humanity (represented in his son whom he must not harm) and a religious obligation to God (represented by his faith that the command to sacrifice his son comes from God).
    There is an implied resolution: we have ethical duties to other human beings only if that duty is grounded in a more foundational and absolute duty to God. Or, more generally: a true obligation dilemma does suggest a problem with an ethical system–there is obligation only if there are absolute duties at the foundation that can outweigh and order the non-foundational duties.
    This is also part of his religious critique of secular ethics: the abstract duty to “humanity”, put into practice, ends up being a duty to multiple individuals, which leads inevitably to conflicts of duties between one individual and another, as well as between individuals and humanity at large. The implication, I think, being that secular ethics fails to found obligation in any absolute duty (e.g., that Kant’s categorical imperative fails). (A further implication might be that the failed attempt leads inevitably to consequentialism, in which the very sense of “obligation” is lost.)

Comments are closed.