Download The Ambiguity of 'Ought' and 'Wrong' and Their Different Senses and more Slides Acting in PDF only on Docsity!
Professor Douglas W. Portmore
OBJECTIVE VS. S UBJECTIVE C ONSEQUENTIALISM AND
ACTUAL VS. EXPECTED UTILITY
I. Why ‘Ought’ and ‘Wrong’ Are Ambiguous Consider the following: Someone might ask: “What ought I to do when I do not know what I ought to do?” (Zimmerman 2006, 359) I might figure out that, given my epistemic position, I ought to do x , and yet I can still wonder whether that’s what I ought to do x. Thus I might ask someone in a superior epistemic position whether I ought to do x , and I am clearly not asking her whether, given my current epistemic position, I ought to do x , for I already know the answer to that question. Ewing says: “We may believe…that the soldiers who fight against us in a war are acting wrongly in fighting, yet every reasonable person will admit that, as long as they really think they ought to fight, they ought ‘to obey their consciences’ and fight” (as quoted in Zimmerman 2006, 335). Ewing seems to be claiming, as many others have claimed, that you ought to do what you think is right. In other words, you ought to do what it is that you believe that you ought to do. Of course, none of the foregoing makes any sense if there is only one sense of ought. Thus we should assume that ‘ought’ and it correlatives (such as ‘wrong’) are ambiguous and have different senses. II. The Different Senses of ‘Ought’ and ‘Wrong’ Below, I’ll attempt to disambiguate the word ‘wrong’, but, corresponding to these different senses of ‘wrong’, there will be different senses for its various correlatives: ‘ought’, ‘obligatory’, and ‘permissible’. To get these different senses of the correlatives of ‘wrong’, we need only to keep in mind the following relations. An act is morally permissible if and only if it is not wrong. And an act is morally obligatory (or, in other words, what the agent morally ought to do) if and only if all of its alternatives are wrong.^1 First, let’s assume that there is a sense of ‘wrong’ for which moral theories are concerned to give criteria. Call this: “‘wrong’ in the fact‐relative sense.” We can use this sense of ‘wrong’ to define two other senses of ‘wrong’. (^1) The following is adapted from Parfit’s Climbing the Mountain. Last Updated: 4/7/08 Page 1 of 13
FRS S’s performing x is wrong in the fact‐relative sense if and only if the correct criterion of rightness along with all the other relevant facts entail that S’s performing x is wrong. MBRS S’s performing x is wrong in the moral‐belief‐relative sense if and only if S believes that her performing x is wrong in the fact‐relative sense. ERS1 S’s performing x is wrong in the evidence‐relative sense if and only if, given the evidence available to S, S ought to believe that her performing x is wrong in the fact‐relative sense. The first two bullets in Section I seem to be trading on an ambiguity between ought in the fact‐relative and evidence‐relative senses. Thus, I might ask: “What I ought to do in the evidence‐relative sense when I do not know what I ought to do in the fact‐ relative sense?” And, as the second bullet suggests, I may know what I ought to do in the evidence‐relative sense, but still wonder whether this is what I ought to do in the fact‐relative sense. The third bullet in Section I, by contrast, seems to be trading on an ambiguity between ought in the fact‐relative and moral‐belief‐relative senses. The thought seems to be that you ought, in the moral‐belief‐relative sense, to do what you believe that you ought to do in the fact‐relative sense. III. The Evidence‐Relative Sense Of these three senses, the evidence‐relative sense is the trickiest. A. Why ERS1 Is Unsatisfactory ERS1 is unsatisfactory, as Parfit’s Mine Shafts example shows: Consider Mine Shafts: A hundred miners are trapped underground, with flood waters rising. We are rescuers on the surface who are trying to save these men. We know that all of these men are in one of two mine shafts, but we don’t know which. There are three flood‐gates that we could close by remote control. The results would be these: The miners are in
is morally permissible and negative if and only if it is morally impermissible. Some immoral acts are morally worse than others. Clearly, this is the case where one immoral act has worse consequences than another. But it may even be that some immoral acts are morally worse than others even though its consequences are no worse than those others. For instance, it might be that intentionally causing harm is morally worse than negligently causing harm even when the consequences in each case are equally bad. Some permissible acts are morally better than others. Clearly, this is the case where one permissible act has better consequences than another. But it may even be that some permissible acts are morally better than others even though its consequences are no better. For instance, it might be that the act of sacrificing an arm to save a life (where this is necessary to save the life) is morally better than the act of sacrificing a $100 to save a life (where this is necessary to save life), and this might be true even if the consequences of the former are worse than the consequences of the latter. c. Expectable Deontic Value Suppose that there are a number of possible deontic values that a given act alternative, a i, may, for all you know, have: DV1(a i ), DV2(a i), DV3(a i),…DVn (a i). Suppose also that for each of these possible deontic values there is a probability, given your epistemic position, that it has this deontic value: Pr[DV 1 (a i)], Pr[DV2(a i)], Pr[DV3(a i)],…Pr[DVn (a i)]. Also, let’s assume that for each act alternative, a i, there is a set of possible future courses of action (FCA) that are available to you if and only if you perform this act: FCA 1 , FCA2, FCA3,…FCAn. Each of these FCAs has an expected deontic value (eDV): eDV(FCA1), eDV(FCA2), eDV(FCA3),…eDV(FCA n ). And, for each of these FCAs, there is a probability that you’ll complete that FCA given your performance of a i: Pr(FCA 1 /a i), Pr(FCA 2 /a i), Pr(FCA 3 /a i),…Pr(FCAn /a i). The expectable deontic value (EDV) of an act, a i, is equal to: ∑(j=1‐n) {DVj(a i) × Pr[DVj(a i)]}
- ∑(j=1‐n) {eDV(FCAj) × Pr(FCAj/a i)}. d. ERS Illustrated Let’s consider a couple of examples.
C ASE 1: Assume that you promised your ex‐girlfriend that you would attend her wedding, but that now you are realizing that the probability that you would complete a future course of action in which you get drunk and punch the groom in the nose were you to accept the wedding invitation is quite high relative to your epistemic situation. Let a 1 be the act of keeping your promise to attend the wedding, and let a 2 be the act of breaking your promise to attend the wedding. Assume that the expectable deontic values are as follows: EDV(a 1) = {(10 × 1) + [(10,000 × .9) + (11,000 × .1)]} = 10, EDV(a 2) = [(₋ 10 × 1) + 10,900] = 10, “10” is the deontic value of your keeping your promise, and “‐10” is the deontic value of your breaking your promise. Assume that if you keep your promise, there is, given your epistemic position, a 90% chance that you’ll undertake a future course of action that involves your getting drunk at the wedding and punching the groom in the nose and that there is only a 10% chance that you’ll undertake a future course of action that involves your behaving yourself at the wedding. The deontic values of these two FCAs are 10,000 and 11,000, respectively. Assume that sum of the expected deontic values of all the possible FCAs that you might undertake if you were to break your promise to attend the wedding is 10,900. Also, assume that, given your epistemic position, you ought to believe that a 1 ‘s expectable deontic value is lower than a 2‘s. This is true even though you obviously don’t have anything close to the precise sort of information about deontic values and probabilities that I give above. In this case, then, you ought, in the evidence‐relative sense, to do what’s wrong in the fact‐relative sense: that is, break your promise (i.e., a 2). In this instance, doing what you know that you ought to do in the fact‐relative sense is just too risky, for there is, given your epistemic position, a very good chance that, if you keep your promise, you’ll do something much worse than breaking a promise: you’ll assault the groom. C ASE 2: Imagine that you are faced with the situation in Parfit’s Mine Shafts example. Suppose, then, that there are three act alternatives available to you, a 1 ‐a 3, corresponding to each of the three gates that you could close. Assume that the expectable deontic values are as follows: EDV(a 1) = [(10,000 × .5) + (0 × .5)] + 9,500 = 14,
This view is either too strong or too weak, depending on whether it requires us to act, or only to try to act, in these ways. On the one hand, if it requires us to act in these ways, then it is too strong, for the evidence as to which acts would make things go expectably best will often be unattainable. On the other hand, if it requires only that we try to act in these ways, then it seems too weak, as we may try to act so as to produce the expectably best outcome while ignoring relevant evidence that we should have taken into account. b. Neutrality Expectabilism seems either to ignore important deontic factors or to assume that the degree to which an act can be morally good/bad is solely a function of the goodness/badness of its consequences. ERS takes all deontic factors into account and is neutral with respect to whether the degree to which an act can be morally good/bad is solely a function of the goodness/badness of its consequences. IV. Which Sense Is Most Salient to Our Moral Deliberations and Assessments of Blameworthiness? A. Which Sense Is Most Salient to Our Moral Deliberations? In our moral deliberations, we should focus on ERS, for, as we saw above, there are many instances, in which avoiding doing what we know to be wrong in the fact‐relative sense would lead us to act in ways that are just too risky. B. Which Sense Is Most Salient to Our Assessments of Blameworthiness? Acting in a way that is wrong in the fact‐relative sense provides neither (1) a necessary nor (2) a sufficient condition for being blameworthy. (1) If I were to close gate 2 in the Mine Shafts case, and if the miners happen, unbeknownst to me, to be in shaft B, then I will have acted rightly in the fact‐relative sense, but I would still be blameworthy for having taken such a big risk. (2) Were I to close gate 3 in the Mine Shafts case, I would be acting wrongly in the fact‐relative sense, but I would not be blameworthy. Acting in a way that is wrong in the moral‐belief‐relative sense provides neither (1) a necessary nor (2) a sufficient condition for being blameworthy. (1) I can be blameworthy for acting in a way that I do not believe to be wrong in the fact‐ relative sense when my belief that what I’m doing is morally right is the result of negligence or culpable ignorance. (2) Were I to close gate 3 in the Mine Shafts case, I would be acting in a way that I believe to be wrong in the fact‐relative
sense, but I would not be blameworthy. Also, consider the case of Huck Finn. He acted in a way that he believed to be wrong in helping a runaway slave to escape, but he doesn’t seem blameworthy. Acting in a way that is wrong in the evidence‐relative sense provides a sufficient condition for being blameworthy, but probably not a necessary condition for being blameworthy. To see that it doesn’t constitute a necessary condition for being blameworthy, consider the case where I break my promise to my ex‐girlfriend to attend her wedding. I act in a way that that’s right in the evidence‐relative sense, but it seems that my ex‐girlfriend can legitimately blame me for breaking my promise, as I could have gone to the wedding and behaved myself. The fact that I have the moral failing of having the disposition to get drunk and be belligerent at such functions does not excuse me from blame for not keeping my promise. V. Subjective and Objective Consequentialism A. The Distinction Objective Consequentialism: S’s performance of x is wrong if and only if there is some available alternative act that would produce more value. Subjective Consequentialism: S’s performance of x is wrong if and only if, given the evidence available to S, S ought to believe that x ’s expectable value is lower than that of some available alternative. B. Which is more plausible? The answer will depend on which sense of ‘wrong’ we’re trying to give an account of. Clearly, objective consequentialism gives the more plausible account of ‘wrong’ in the fact‐relative sense. And, clearly, subjective consequentialism gives the moral plausible account of ‘wrong’ in the evidence‐ relative sense. VI. Various Versions of the Epistemic Objection to Objective Consequentialism A. The Standard Version of the Epistemic Objection If objective consequentialism is true, then nearly 100% of the acts that people perform are morally wrong. To see this, consider the following sort of cases: (1) Feldman’s safe‐cracking case, (2) Miller’s choking‐neighbor case, (3) Howard‐ Snyder’s beating‐Karpov‐at‐chess case, (4) my talking‐to‐John‐after‐class‐ giving‐rise‐to‐the‐next‐Hitler case, and (5) Wiland’s the proverbial‐monkey‐at‐
iii. Beating Karpov is, of all the acts that she can perform, the one that would produce the best consequences. iv. Therefore, if objective consequentialism is true, then she ought to beat Karpov. But, as Moore points out, (III) and (ii) contradict each other, and Howard‐ Snyder cannot have it both ways. She cannot say that she both can and cannot beat Karpov in the same sense. It seems important, then, to disambiguate two senses of ‘can’, for it seems that in one sense Howard‐Snyder can beat Karpov and in another sense she cannot. Now Moore suggests that the sense in which she can beat Karpov is the sense in which it is physically possible for her to do so. He says, “Her beating Karpov is consistent with the physical laws of nature. Call this sense, ‘can1’” (2007, 87). But this seems to conflict with determinism. If determinism is true, then it may well be that her beating Karpov is inconsistent with the physical laws of nature. It is preferable, I think, to remain neutral as to whether the sense of ‘can’ under which she can beat Karpov is compatible with determinism. So let us say, then, that there is a sense of ‘can’ under which Howard‐Snyder can beat Karpov, Tiger Woods can make a hole‐in‐one, and I can publish in the Philosophical Review even though, in none of these cases, is this something that is under the agent’s control in the sense explained in (I) above. Let’s call this sense, ‘can 1’, and let’s call the sense of can given in (I) above, ‘can 2 ’. It’s clear, then, that if the foregoing interpretation of Howard‐Snyder is correct, then Howard‐Snyder is illegitimately equivocating between these two senses of can. Perhaps, though, the following is a more charitable interpretation of the argument. This is the interpretation that was suggested to Moore by an anonymous referee (viz., me): a. If objective consequentialism is true, then (since Howard‐Snyder can 1 beat Karpov) she ought to beat him. b. But Howard‐Snyder cannot 2 beat Karpov. c. ‘Ought’ implies ‘can2’: if an agent cannot 2 do something, then it’s not true that she ought to do it. b. [ sic ] Therefore, it’s not true that Howard‐Snyder ought to beat Karpov. [from (b) and (c)] e. Therefore, objective consequentialism is false. [from (a) and (d)] (Moore 2007, 87) This argument doesn’t commit the fallacy of equivocation, so it is valid. But, as Moore argues, we should reject (c). If objective consequentialism is to be
formulated in terms of can 1 , then so should the ought‐implies‐can principle. Arguably, ‘can 1’, not ‘can 2’, is the ‘can’ of ordinary language, and we should formulate the ought‐implies‐can principle in term of the ‘can’ of ordinary language, for the ought‐implies‐can principle is justified by appeal to ordinary language. We would not say that when Tiger Woods sinks a hole‐in‐one, he does something he cannot do. Yet sinking a hole‐in‐one is not something he can 2 do. In conclusion, we should reject the notion that objective consequentialism conflicts with the ought‐implies‐can principle. C. Howard‐Snyder Argument in Terms of Blameworthiness and Miller’s Two Arguments
- Howard‐Snyder’s Argument in Terms of Blameworthiness There is one more way we might interpret Howard‐Snyder’s argument against objective consequentialism: a. If objective consequentialism is true, then we are morally blameworthy for performing actions that we can 1 but cannot 2 do. b. But it is not the case that we are morally blameworthy for performing actions that we can 1 but cannot 2 do. c. Thus, objective consequentialism is not true. (Moore 2007, 89) E VALUATING THE ARGUMENT: Premise (a) is false. Objective consequentialism says nothing about when agents are morally blameworthy. Of course, Howard‐Snyder might claim that there is a conceptual connection between wrongness and blameworthiness, but if there is such a connection, the connection is between blameworthiness and ‘wrong’ in the evidence‐relative sense, whereas objective consequentialism gives us a criterion for wrongness in the fact‐relative sense.
- Miller’s First Argument Imagine that while I’m having my breakfast, my next‐door neighbor is, unbeknownst to me, choking. It is possible for me to run next door, break into his house, and perform the Heimlich maneuver on him, thereby saving his life. But since I have absolutely no reason to think that he is choking, the thought of my doing so never even crosses my mind. And so he dies. While saving him is something I can^1 do,^ it^ is^ not^ something^ that^ I^ can^2 do. So consider the following argument:
- Distinguishing between a Criterion of Rightness and a Decision Procedure
- Partners in Guilt
