As I noted in an earlier post, there are at least two competing rationales for constraints: the victim-focused rationale (VFR) and the agent-focused rationale (AFR). One worry that I have about VFR is that it seems to have a tough time accounting for the intuition that I and many others have in the Promise Case. Another worry that I have is that it doesn’t seem to be able to account for the intuition that I and many others have in Case C.
In Case C, Carl takes a pill that causes him to conceive a son (whom he names Oliver) who lives only 40 good years rather than a daughter (whom he would have named Patricia) who would have lived 80 good years. He does this so as to prevent five others from taking the pill. Of those survey-takers who both find constraints intuitive and think that taking such a pill without justification is wrong (these, I take it, are those who didn’t answer “It is not the case that both Abe and Bert acted impermissibly” in response to this survey), 89 percent thought that what Carl did was wrong. (These were the results as of 11:30 a.m. EST.) These people, as I do, have the intuition that there is a constraint against taking such a pill. But I don’t see how VFR can account for this intuition, for there is no victim whose rights have been violated in this case. Surely, Oliver doesn’t have a right not to be conceived just because those who conceived him could have conceived someone who would have lived a better life. And note that Oliver is not made worse off as a result of what Carl does. Had Carl not taken the pill, Oliver would have never been conceived. So I don’t see how Oliver can be said to be the victim. Carl may have been made worse off as a result of his taking the pill, but this needn’t be the case. And surely we don’t think that Carl’s rights are infringed by his decision to take the pill so as to prevent five others from taking the pill. In any case, I think that the intuition that what Carl did was wrong persists even after we stipulate that Carl is made no worse off as a result of his taking the pill. So I don’t see how we can say that Carl is the victim. And I don’t see how we can say that Patricia’s rights are infringed, for she doesn’t exist and the nonexistent have no rights. So Patricia can’t be the victim. Indeed, there doesn’t seem to be a victim at all. And, since there’s no victim, I don’t see how the rationale for this constraint can be victim focused. Of course, just because VFR can’t account for something doesn’t mean that it’s false, but it does seem to indicate that AFR has a leg up on VFR in that it is able to account for everything that VFR can account for and more.
Doug,
Just a comment on your last sentence. Without hoping to show that VFR can account for cases like these, I’d want to say that VFR can do at least one other thing that AFR cannot do: accommodate a very strong theoretical (as opposed to possible-case) intuition that very many people have, namely that in at least many cases, what fundamentally makes the act wrong is something about the patient’s moral status. (VFR just is a version of that, I’d guess.) I am assuming here that AFR and VFR are supposed to be accounts of fundamental wrong- (or right-) makers. And, while I haven’t taken a poll, I take the widespread intuitiveness of that kind of theoretical intuition to be evidenced by, for example, the fact that so many (both in and out of academia) find something like Kant’s Formula of Humanity compelling, or that so many find Parfit’s rejection of person-affecting views worth discussing. If that’s right, then in tallying up the wins and losses of VFR and AFR, we should give VFR at least one point that AFR doesn’t get.
Josh,
But AFR, and not VFR, is compatible with other theoretical intuitions: those that drive people to accept theories such as act-consequentialism, rule consequentialism, contractualism, and so forth.
So I agree that Kantians who take Kant’s Formula of Humanity to be the most fundamental moral principle should give VFR one point that AFR doesn’t get. But those of us who find other types of moral theories attractive given different theoretical intuitions should not give VFR that point.
I think that the issue of which theory starts from the most attractive theoretical intuitions is a whole different debate than the one that I’m trying to have. The debate that I’m interested in having here is whether it is AFR or VFR that best accords with our first-order moral intuitions.
Doug,
I’m not sure that VFR isn’t compatible with those other intuitions. (I’m not sure, in part because you don’t specify those other intuitions, and in part because I haven’t tried to work it all the way out. But, for example, I would have suspected that VFR is compatible with contractualism, at least as some folks understand it.) But in any case, I think I’ve lost the track of your argument here. I thought your argument was something like this: many people — not everyone, but many people — have a certain intuition about Case C, which AFR, but not VFR, can account for, which means, roughly, that AFR has a point of plausibility that VFR does not share. My argument was meant to be parallel to that: many people — not everyone, but many people — have a certain intuition about the moral status of patients, which VFR, but not AFR can account for…. It now sounds like you’re saying something else: people who have (what I’ll call) the AFR Intuition in Case C should reject VFR, unless some equally or more compelling factor pushes them to abandon the AFR Intuition. But then it overstates things to say that AFR has a leg up on being able to “account for everything.” And, a parallel argument is available to fans of VFR: people who have (what I’ll call) the Patient-Focused Theoretical Intuition should reject the AFR Intuition in Case C, unless some equally or more compelling factor pushes them to abandon the Patient-Focused Theoretical Intuition.
This is all just a dialectical question, I guess. But it seems to me that people with the Patient-Focused Theoretical Intuition should give their opponents a point in regards to Case C. Equally, then, if you grant the evidence I cited earlier, people with the AFR Intuition in Case C should give their opponents a point in regards to theoretical intuitions. I don’t think these are different debates: I see the question as which moral theory can ultimately accommodate the many, sometimes conflicting, moral judgments we make, where those judgments include not only possible case intuitions but also theoretical intuitions. But again, this is just a bigger-picture, dialectical point. You could, in principle, give me all of this and just make your point by revising your last sentence to say something like, “…AFR has a leg up on VFR in that it is able to account for every possible case intuition that VFR can account for and more.” (Assuming there are no possible case intuitions that VFR can uniquely accommodate.)
Josh,
I would have suspected that VFR is compatible with contractualism.
But, on contractualism, what fundamentally makes the act of, say, promise-breaking wrong is that it violates a principle that cannot be reasonably rejected, not fundamentally something about the patient’s moral status. Right? Or am I not understanding what you mean when you say that moral status is fundamentally what makes these sorts of acts wrong?
When I said that AFR can “account for everything” that VFR can account for, I was speaking of first-order moral intuitions — intuitions about particular cases. So you’re suggested revision of my last sentence is exactly what I meant to have said.
But it seems to me that people with the Patient-Focused Theoretical Intuition should give their opponents a point in regards to Case C. Equally, then, if you grant the evidence I cited earlier, people with the AFR Intuition in Case C should give their opponents a point in regards to theoretical intuitions.
I think that only those people who share my intuition with regard to Case C should give a plausibility point to AFR. And I think that only those people who share your theoretical intuitions should give VFR a plausibility point. I don’t think anyone should give a view more plausibility points merely because many other people have an intuition that only that view can account for. What seems plausible to one person may not seem plausible to another, and the mere fact that others find something plausible doesn’t necessarily make it seem any more plausible to oneself.
Since I don’t share your theoretical intuitions, I don’t think that I should give VFR any plausibility points. You should, though.
Okay, so it sounds like we’re basically on the same page.
Two side-notes. First, what I had in mind with contractualism was that Scanlon at multiple points says something along the lines of, “justifiability to others is close to being equivalent with or based on a Kantian-style principle of the value of rational nature.” I haven’t really tracked the literature on this question, but I think some folks have questioned whether Scanlon really therefore counts as a contractualist. For those folks, contractualism and VFR would not be compatible. But if Scanlon gets to count as a contractualist (!), then assuming he’s not wrong about the compatibility, contractualism and VFR are compatible. In that case, then, contractualism might not count as a distinctive theory of what fundamentally makes acts right or wrong.
Second, I don’t want to hijack your post with this, but I’d be curious to know how you see things when a philosopher consciously bites a bullet. I always thought there were sort of two ways of doing this: (1) I accept that my view has an implausible implication, but that is the price of consistency, and everything else about my view is awesome, or (2) I accept that everyone else will think that my view has an implausible implication, but I don’t agree, and that is the price of consistency, and everything else about my view is awesome. It sounds like you don’t think 2 should count as a flaw of the view, as a bullet to be bit. I do, I guess; if 2 is true of a view, it doesn’t mean that it’s false, but it’s a downside nonetheless.
Josh,
I don’t think that the fact that everyone else believes ~q is, in and of itself, any reason for me to believe ~p. And that’s true even if I know that p entails q. If I have some reason to think that it’s less likely that everyone could be wrong in thinking that ~q than that I could be right in thinking that p, then this fact in conjunction with the fact that p entails q gives me some reason to believe ~p.
The downside to not sharing other people’s beliefs or intuitions is not that that fact should, in and of itself, make you less confident in your own beliefs or intuitions, but that it makes convincing other people more difficult.
The point of the surveys was not to determine whether I should be more or less confident in my own intuitions, but to determine whether I could use them as premises in arguments and expect anyone to accept those premises.
Doug
I am not clear on why you think that what Carl did was not permissible. What is the constraint that suggests that what Carl did was not permissible? Other then the intuition that it is impermissible to take the pill, what reason do we have that warrants us using this intuition? Or is having the intuition that Carl taking the pill is impermissible sufficient to claim that Carl taking the pill is impermissible? If this is what you are suggesting do we need an error theory to account for the intuitions of the other 11%?
Am I missing something?
John,
By definition, there is a constraint against performing a certain act-type if and only if agents are prohibited from performing that act-type even in certain instances in which performing that act-type would minimize overall commissions of that act-type. This means that if Carl’s act is impermissible, then there is a constraint against taking that sort of pill.
But now you want to ask, “Why do I think that Carl’s act is impermissible?” Answer: Because I think that it is, absent a sufficiently good reason, wrong to conceive children that will have substantially worse lives than the children that one could otherwise conceive. And I don’t think that preventing five others from doing the same is a sufficiently good reason for doing so. I know that you probably want more. But try asking yourself: “Why do you think that it is wrong to cause gratuitous suffering even in order to prevent five others from doing the same?” You can, I hope, see that it’s difficult to answer such questions.
In any case, the reason that you don’t find any defense of my intuition with regard to Case C is that it was not my intent to do so. My intent was only to argue that if you have the same intuitions that I have (and as the survey shows, many of my readers do), then you have some reason to favor AFR over VFR.
Doug,
I’m just going to belabor the point a bit. If you want to skip it to go back to stuff more directly related to your post, no worries. Okay, that said, my naive armchair-epistemology might not quite go the same way you are going, but that’s not what I was trying to say — I didn’t take you to be trying to fix on the confidence you should have in your intuitions. I took you to be defending AFR. My usual way of looking at these things is that if a defense of some thesis cannot accommodate something everyone believes, that is a strike against the defense. I guess a lot of this depends on what an argument is supposed to do (who it’s supposed to convince, in particular), but that’s all I meant to say. (Of course, not everyone believes the kind of theoretical intuitions that support VFR, but very many do.)
Doug,
I understand that you don’t wish to offer a full justification for your intuition in this case; I would be hard-pressed to do so myself. However, I wonder whether whatever justification is ultimately on offer won’t end up protecting VFR from your objection. If there really is no victim, and what was done is worse for no one and nothing, I think many would find it hard to ever justify the intuition that what Carl did was wrong. Mightn’t it be necessary to find a way to talk about taking the pill as though it is not victimless, even if the victim in question is not a “definite” entity? I think it plausible that we want future persons to be as well-off as possible, whomever they may be. It is in the cashing out of this final clause that the difficulty lies, but it may be that it is possible to explain this intuition by recognizing as a moral patient whoever fills some “future slot” (or something similar, or even completely different). I realize that this is very, very, very rough, but if something like this is needed (as I think it is) to explain the intuition that what Carl does is wrong in the first place, then the ultimate solution seems likely to work just as well to rescue VFR from your objection.
Doug
Thanks for clarifying what you mean by ‘constraint.’ I was not looking for more. I am trying to understand the discussion within the context you set up. I am not sure that I agree with the claim you are making about the impermissiblity fo performing an action even if performing it will minimize other instances of it from happening, but that is for another discussion. I do understand how complex and difficult ethics/philosophy can be. I have been studying philosophy for nearly 45 years and teaching for over 20.
I do have one other question, if we have a reason to favor AFR over VFR in these types of cases then do we have a reason to think that VFR is wrong in general? Or is there room for favoring one over the other in some instances, but realizing that the other may have merit in understanding other types of cases? Ot, it is possible that they collapse into another FR? I know this is outside the intent of your post, but if you are willing I would appreciate an answer. I will not ask others that go beyond the stated intent of your post(s).
Josh,
a defense of some thesis cannot accommodate something everyone believes, that is a strike against the defense. Is there a distinction between there being a strike against the defense of some thesis and there being a strike against the thesis? And what do you mean by a “there being a strike against x”?
I admit that an argument for a thesis that employs premises that others don’t accept won’t convince those others. I don’t think, though, that this means that the argument is unsound. So if by “defense of x”, you mean “an argument for x,” and if by “a strike against the defense of x,” you mean counts in favor of thinking that the argument is unsound, then I disagree. But if all you’re saying is that a defense that rests on assumptions that others don’t accept is going to be unpersuasive, then I agree.
David,
I don’t understand your proposal. Who’s the future victim? When you talk about the moral patient, whoever he or she is, who fills some “future slot,” who is that in Case C? Oliver? Patricia? Some indeterminate person? And if the last of these, what indeterminate person?
John A.,
if we have a reason to favor AFR over VFR, then we have some reason to think that AFR offers a better explanation for typical constraints (like those against murder and promise breaking) than VFR does.
It’s entirely possible that there are other reasons to favor VFR over AFR. But I haven’t been able to think of any. Josh Glasgow suggests one, but I don’t find his reason compelling since I don’t share his intuition. I share his intuition that moral status is important, just not his intuition that the fundamental wrong-making feature of actions is a failure to respect humanity.
It’s possible that there is some other competing rationale that is better than both AFR and VFR, but I can’t think of what that might be.
Doug,
The idea is that there can be de dicto victims. Carl has done wrong because he has harmed his future child, whoever that turns out to be, and so the victim of his action is ‘Carl’s future child’ (de dicto). This is (I take it) the view Caspar Hare defends in this paper.
In any case, I don’t think Hare’s need be the correct view to make my point: Many people accept some version of the person-affecting principle (PAP). There is an apparent tension between PAP and the intuition that Carl has done something wrong. So, one must either give up the intuition, give up PAP, or find some way to relieve the tension. My hypothesis is that any viable tension-relieving solution (such as Hare’s) would serve just as well to protect VFR from your objection. If I’m right, then it seems unlikely that Case C will convince anyone committed to PAP to give up VFR.
I wonder if we have to choose between AFR and VFR. It seems to be that both can be right. It could well be that some constraints are created by the qualities of the victims whilst on other occasions features of our agency constrain what we should do. It could well be that in these non-identity cases it is the qualities of our agency that do the work. But I don’t think it is a good argument to claim that therefore all constraints must be created by the features of agential involvement. This seems to go the other way round too. In the killing someone in order to prevent other killings, it does seem plausible that the moral status of the victim creates the constraint but again there is no reason to make the induction that this must always be the case. So, I guess I would be happy with AFR being right about some cases and VFR about others.
David,
I now understand your point and accept it. I had always thought that the conflict between cases like Case C and PAP was actual, and not merely apparent. But it’s good to know that others disagree and that there are arguments to the contrary.
Jussi,
I wonder if we have to choose between AFR and VFR. It seems to be that both can be right. It could well be that some constraints are created by the qualities of the victims whilst on other occasions features of our agency constrain what we should do.
I agree.
It could well be that in these non-identity cases it is the qualities of our agency that do the work. But I don’t think it is a good argument to claim that therefore all constraints must be created by the features of agential involvement.
Contrary to what you’re suggesting, the argument isn’t:
(1) AFR, but not VFR, can explain the constraint in Case C.
(2) Therefore, AFR and not VFR.
The argument is rather:
(1) It seems that AFR can account for every case intuition that VFR can account for and some that VFR cannot account for. Thus, AFR has greater explanatory power.
(2) Other things being equal, the better explanation is the one with greater explanatory power.
(3) Therefore, other things being equal, AFR is the better explanation.
Now, one could argue, as Josh has, that other things are not equal — that VFR, and not AFR, is compatible with certain theoretical intuitions. And one could argue that there are certain case intuitions that VFR, but not AFR, can account for. But I haven’t heard anyone give such a case yet. I would be eager to hear if others thought that there are any, and, if so, what they are.
In the killing someone in order to prevent other killings, it does seem plausible that the moral status of the victim creates the constraint.
The issue is not whether the moral status (e.g., the rights) of the victim is what gives rise to the constraint, for the advocate of AFR can accept that it does. Both the advocate of AFR and the advocate of VFR can agree that with regard to many typical constraints the rights of the victim give rise to the constraint. They disagree as to the nature of that right. The advocate of VFR thinks that that right acts as a barrier against the permissibility of infringing on that right even so as to minimize one’s own infringements of that right. (Infringements contrast with violations of a right in that only the latter is necessarily wrong.) The advocate of AFR disagrees and thinks that it is permissible to infringe on that right so as to minimize one’s own infringements of that right. In other words, then, the issue is whether it is ever permissible for A1 to violate C1 at T1 even if this is the only way (short of doing something even worse) for A1 to prevent herself from violating C1 on each of five separate future occasions: T2-T6. VFR predicts no, and AFR predicts yes. The problem is that it’s very hard to come up with a test case when the constraint is one that involves murder. Typically, there are better ways of avoiding voluntarily committing numerous murders in the future than by voluntarily committing one murder now: namely, by just choosing in the future not to voluntarily commit murder, or by locking yourself up. In other words, it is relatively easy to come up with cases where one must break either one promise now or five promises in the future and where each of these infringements will be comparable, but it’s very difficult to come with cases where one must commit either one murder now or five murders in the future and where each of these infringements will be comparable. Kamm’s Bomb Case is an attempt to do the latter, but, as I argued in Part I, it’s a poor test case, because the instances are not comparable and there’s some question as to whether set of constraints infringed is the same in this case.
Everyone,
Given all the helpful discussion, this is how I would now set up the distinction between AFR and VFR (while getting rid of the potentially misleading labels that proponents of VFR have imposed):
There are two possible explanations for the judgments entailed by typical sorts of constraints (e.g., the constraint against harming innocent others and the constraint against breaking a promise):
E1: The explanation lies with the fact that the victim has a certain sort of right (a constraining right), one that acts as a barrier against the permissibility of infringing on that right even for the sake of minimizing infringements of that right by oneself or by others.
E2: The explanation lies with some fact other than the one stated in E1 — such as, perhaps, the fact that agents bear a special responsibility for what they do as opposed to what they allow to happen.
Right. Now I am a bit confused. I’m afraid that there’s some talking past one another going on. In the first post you said that the question is over what explains (1). Now you say that both views can agree that this is something about the status of the victim. The new debate seems to be about the nature of rights. I agree that this might be an interesting question to ask. But it doesn’t seem like a question about the rationales for the constraints.
Here’s a case. I really really want to go the pub tonight. If I go the pub, I will get drunk. If I get drunk, I’m going to hit at least five people. Because I am so weak willed, the only way to stop myself from doing this is to hit a bystander now so that I will be locked up tonight. Is it permissible to punch the bystander?
Jussi,
I think that the debate is essentially the same as before, although clearly my views have evolved in light of the discussion. The debate, in essence, is whether or not philosophers such as Frances Kamm are right in thinking that the best explanation for intuitions such as 1A and 1B (see below) is E1 (see above).
1A: It would be impermissible for an agent, A1, to commit murder even if this is the only way (short of doing something even worse) that A1 could prevent A2-A6 from each committing a comparable instance of murder.
1B: It would be impermissible for an agent, A1, to commit a promise-breaking even if this is the only way (short of doing something even worse) that A1 could prevent A2-A6 from each committing a comparable instance of a promise-breaking.
E1 is just a clearer statement of VFR. And the issue is and always has been whether E1 offers the best explanation for intuitions such as 1A and 1B. I’ve argued that it doesn’t, because there are versions of E2 that, unlike E1, get the right answer in the Promise Case. Unlike E1, E2 predicts correctly that 3A (see below) is false.
3A: It would be impermissible for A1 to commit a promise-breaking act at T1 even if this is the only way (short of doing something even worse) that A1 could prevent herself from committing comparable promise-breaking acts on each of five separate future occasions: T2-T6.
Also, I think that E2 has greater explanatory power in that, unlike E1, it can account for 1C.
1C: It would be impermissible for an agent, A1, to commit a sperm-mutating pill-taking act even if this is the only way (short of doing something even worse) that A1 could prevent A2-A6 from each committing a comparable instance of a sperm-mutating pill-taking act.
Now to the case that putatively E2 cannot explain. You say,
Now, I have to ask two questions. First, is your intuition 3B or ~3B. My intuition is ~3B.
3B: It would be impermissible for A1 to commit a hitting act at T1 even if this is the only way (short of doing something even worse) that A1 could prevent herself from committing comparable hitting acts on each of five separate future occasions: T2-T6.
Second, I have to ask: Could you both not hit the bystander and not hit the five others? I realize that you will hit the five others if you don’t hit the bystander, but I’m asking whether you could not hit the five others even if you don’t hit the bystander. If that’s something you could do, then you haven’t given us a test case for 3B or ~3B. A proper test case is one where the only way that A1 could prevent herself from performing x five times in the future is to perform x once now.
But perhaps you are indeed supposing that if you don’t hit the bystander, then you lose control of your actions such that you cannot both not hit the bystander and not hit the five others. In that case, it’s true that the only way you could prevent yourself from hitting the five others is to hit the bystander. But this still isn’t a proper test case for 3B or ~3B. The present and future instances of hitting are not comparable. The first instance is voluntary, whereas the latter five are in involuntary.
So I don’t think that you’ve given an adequate test case for determining whether we have intuition 3B or not. But suppose that you could devise one and that our intuition is 3B. E2 could account for this intuition, although it would need to appeal to something like the fact (assuming that it is a fact) that we bear a special responsibility for our present actions.
In any case, do you see why I think that it’s so tricky to come up with test cases regarding 3-type intuitions when the constraint involves a restriction (against, say, hitting) as opposed to a special obligation (to, say, keep our promises)?
Doug,
I don’t think we’re very far apart anymore. The answers to your questions are, respectively: yes and something like “a flaw in.” (And if you ask for more than “a flaw in,” I probably won’t be able to give it to you.) But by “defense” I don’t mean just an argument; I mean something that’s supposed to convince people, too, in particular by having premises that people are likely to find compelling, all things in. That’s a higher demand than mere soundness, obviously. But again, I don’t think we’re really disagreeing anymore. Thanks for the discussion!
Doug
“By definition, there is a constraint against performing a certain act-type if and only if agents are prohibited from performing that act-type even in certain instances in which performing that act-type would minimize overall commissions of that act-type.”
I am wondering if the number of people positively affected is a factor in which intuition is generated? Would the intuition change if by Carl taking the pill 1000 other men would not? I take it that there should be no such number given how you have defined ‘constraint.’ Or, are intuitions context dependent such that there might not be a constraint when the numbers reach a certain threshold generating the intuition that what Carl did is permissible, while the constraint would be in effect in cases below that threshold.
John,
I am wondering if the number of people positively affected is a factor in which intuition is generated? Would the intuition change if by Carl taking the pill 1000 other men would not?
Yes. The constraint is, I believe, non-absolute.
I take it that there should be no such number given how you have defined ‘constraint.’
Wrong, but perhaps my definition was a bit unclear. When I said that “agents are prohibited from performing that act-type even in certain instances in which performing that act-type would minimize overall commissions of that act-type,” this means that agents are prohibited from performing that act-type even in some (not all) instances in which performing that act-type would minimize overall commissions of that act-type.
Doug
If I understand you then I could use ACR to justify Carl taking the pill if 5 is within my threshold number. I am wondering then if the intuitions that are being generated by these experiments are not related to the numbers who are affected rather then the acceptability of ACR or VCR.
Thanks for a series of enjoyable posts. I do hope you have more.
John,
I’m not too worried about this possibility. Since 88% of the relevant respondents had the intuition that Carl did wrong in Case C, they presumably think that whatever threshold this constraint may have is not met when only five preventions are at stake. But if your point is that even more than 88% might have had the intuition that what Carl did was wrong had I changed Case C so that Carl’s taking the pill would prevent only two (as opposed to five) other men from doing the same, then I agree.
Doug
But would less have answered that Carl did something impermissible if the thought experiment stated a larger number (1000) not taking the pill as a result of Carl taking it? The underlying problem for me is that the threshold number would be arbitrary to the person setting it and that it places us in a ‘paradox of the heap’ situation. If it is permissible for Carl to take the pill if 1000 others will not, why not 999, etc. Was the experiment biased due to the numbers involved?
Thank you for taking the time to continue the discussion.
Had Carl not taken the pill, Oliver would have never been conceived. So I don’t see how Oliver can be said to be the victim.
I’m late to the discussion, Doug, but this is very interesting. I’m not sure why you believe that Oliver could not have been a victim. Let’s stipulate that Oliver might have been conceived, in the case where Carl does not take the pill, but it occurs in a pretty distant world. So the might counterfactual is not true. Now Oliver should put some credence (i.e. some probablity) in the proposition “Carl does not take the pill and I am conceived anyway”. How valuable is the world in which that proposition is true? It might be extremely valuable to Oliver. He might say, “If my choice is between the gamble: G= [p, I exist 80 yrs; 1-p, I do not exist at all] and the option in which I exist for sure and live 40 yrs, I much prefer the gamble. The value of the gamble might be much higher to him than the pure strategy; I don’t think that is impossible (in fact I think this might explain the resentment in certain versions of the handicapped child case). If so, then Carl did harm him, I think, since Oliver prefers that the gamble have been taken. Carl acted in such a way that a very important preference of Oliver’s is now frustrated.
Mike,
You suggest that we “stipulate that Oliver might have been conceived, in the case where Carl does not take the pill.”
We could do that, but why would we want to do that? Is doing so going to help us test whether E1 or E2 offers a better explanation for many typical constraints? I don’t think so.
In Case C, I stipulated that if Carl doesn’t take the pill, he conceives a girl, whom he names Patricia. Do you think that Oliver could have been a girl named Patricia? In any case, what’s wrong with me just stipulating, as I did, that if Carl takes the pill, Oliver is not conceived?
John,
The issue of how do we determine a non-arbitrary threshold for non-absolute constraints seems orthogonal to the one that I’m concerned with here: viz., whether it is E1 or E2 that provides a better explanation for intuitions such as 1A and 1B.
In Case C, I stipulated that if Carl doesn’t take the pill, he conceives a girl, whom he names Patricia. Do you think that Oliver could have been a girl named Patricia?
I’m not sure whether Oliver could have been a girl (well, I do have a view about this, but I don’t want to derail). All I’m suggesting is that there is some world in which Carl does not take the pill and Oliver is born anyway. Certainly there is such a world, right? I mean, Carl might have frozen the sperm that matter here and so on. All I need is the possibility, and then I get the non-zero credence for that world. And if so, then if the value of that world is high enough, we have a case in which, plausibly (or, so say I) Carl has actually harmed Oliver by taking the pill. So my point was about the harm question.
Mike, that’s really a good point. Hm.
Here are a couple of questions. First, suppose Jen has a birth defect, something significant but not horrible. Her mother could have taken a medication that would have prevented the birth defect but reduced the chance that Jen would be conceived. So, had her mother taken the drug, the prospect for Jen would have been: .5 a better life than she actually has; .5 no life at all. The question is: how are we supposed to evaluate the value of this prospect to Jen?
That question reminds me of John Broome.
The second question is this. In Doug’s scenario, this seems true:
But this would conditional seems to contradict the might conditional,
Can you convince us that the would conditional is false? Or do you think the would conditional is consistent with the might conditional?
That question reminds me of Al Hájek.
Damn, I didn’t notice that I forgot to close off my italics. Sorry.
Hi Jamie,
The question is: how are we supposed to evaluate the value of this prospect to Jen?
How about this? Jen now has a very strong preference that her mother had taken the gamble. That preference is frustrated by her mother’s having failed to take that gamble, and would not have been frustrated had she taken it. So Jen’s mother acted in a way that now frustrates a very strong preference of Jen.
Can you convince us that the would conditional is false?
Right, I don’t want to argue that the would conditional is false. So I agree that,
1. ~P []-> ~O
But (1) is consistent with (2),
2. <>(P & O)
(1) is consistent with (2) so long as the world in which P & O are true is not among the closer worlds.
Hm, now I can’t close off the italics. Anybody fluent enough in TypePad to know how to fix this?
I agree that Jen’s mother acted in a way that frustrates a strong preference of Jen’s (now — that is, a pref. that she now has). But this doesn’t yield immediate consequences regarding how good or bad the action was for Jen.
My point about the conditionals was this: most people think that
“If A, then it would have been the case that B”
is at least the contrary (and maybe the contradictory) of
“If A, then it might have been the case that not-B”
Do you mean to be denying this conditional-contrariness principle?
Jamie,
About the conditionals, I meant to say that these two are consistent.
1. ~C []-> ~O
and
2. <>(~C & O)
So, despite the conditional in (1) being true, it is certainly possible (in this case and in most cases) that ~C occurs and O occurs. But that’s ok, since (1) is not a strict conditional. Of course in that (~C & O)-world, (1) is false. So, I want to assert both (1) and (2). But (2) does not entail that the might conditional in (3) is true. Indeed, it is false.
3. ~C <>-> O
I agree that Jen’s mother acted in a way that frustrates a strong preference of Jen’s (now — that is, a pref. that she now has). But this doesn’t yield immediate consequences regarding how good or bad the action was for Jen.
I think I want to say that Jen might have a legitimate grievance against her mother. At the time of the action, the option available to her mother was to take a gamble having a partitioned outcome. The outcome is O= [(.5) Jen is much better off; .5 Jen does not exist]. But Jen might urge that O is better than O’ = Jen exists with a serious disability. Jen might say that partitioned outcome is clearly the better outcome for me. We can imagine that Jen prefers to take a new drug which has just that partitioned outcome to taking no drug at all.
The italics should be good now.
(Hope this is not a double post, if so, apologies)
Jamie,
I’m asserting only that (1) and (2) are consistent.
(1) ~C []-> ~O
(2) <>(~C & O)
The consistency results from (1) not being a strict conditional. For the record, I agree that (3) is true.
(3) ~(~C <>-> O)
(2) gives me all I need, since it allows me to place some credence in the occurrence of the outcome (~C & O).
I agree that Jen’s mother acted in a way that frustrates a strong preference of Jen’s (now — that is, a pref. that she now has). But this doesn’t yield immediate consequences regarding how good or bad the action was for Jen.
Jen might argue this way. There are two possible outcomes of her mother’s action. One outcome is a partition, O
O = [.5, Jen’s life is very good;.5, Jen does not exist]
The other outcome is O’.
O’ = Jen lives with a serious disability.
But from Jen’s point of view, O > O’. So, Jen’s mother acted then in a way that harms Jen now.
Thanks Josh.
But Mike, if the might conditional is false, then how what is Oliver’s complaint? He can’t truly say, “If only you hadn’t taken the pill I might have been much better off.”
And for Jen: but how is she supposed to evaluate that lottery prospect? On what grounds does she say that O > O’?
Here’s what Oliver can say. It’s possible that you do not take the pill and I’m alive anyway. That is, it’s not necessary that, if you don’t take the pill, I am not alive. Now, you might say, well, the chances aren’t good that you are alive, if I do not take the pill. I say, I agree, the chances aren’t good, but they’re not zero and they’re not infinitessimal. If I really value that world in which you do not take the pill and I live, then I might rationally prefer to have taken the chance. That is, I might rationally prefer that you had not taken the pill.
This could be true because the value of the gamble G = [p, I live & no pill; 1-p, I do not live & no pill] might be greater than the value of the world I’m in, (I live & pill). It all depends on how valuable it is to me that my lifetime is doubled.
Does that make better sense?
Mike,
Are the following true? It’s possible that Carl does not take the pill and Oliver lives 80 miserable years. The chances of this are not zero. The value of the gamble G’ = [p, Oliver lives 80 miserable years & no pill; 1-p, Oliver doesn’t live & no pill] might be lesser for Oliver than the value of the world that Oliver is in.
I think this might well be true. So what you’re pointing out is that the partition ought ot be made finer. How many worlds are there in which he lives to 80yrs and does well and how many in which this does not happen. Take that partition of worlds (the worlds in which he lives to 80 or beyond, some good and some bad) and generate a gamble over those worlds. The value of that gamble (I’m suggesting) might be higher than the value of the world he is actually in. It might be worth the risk to take the gamble.
Mike,
The value of that gamble (I’m suggesting) might be higher than the value of the world he is actually in. It might be worth the risk to take the gamble.
And it might not.
In any case, given that I specified, in my description of the case, that everything else is equal and that I didn’t leave out any morally relevant details, I think that the fair thing to do is to assume that the disvalue of all bad-outcome gambles balances out the value of all good-outcome gambles. Moreover, couldn’t I just stipulate that this is so? Or are you worried that the responses to the survey would have been significantly different had I made this stipulation more explicit? Myself, I doubt that very many people were thinking: “It’s possible that Carl does not take the pill and Oliver lives 80 good years. The chances of this are not zero. The value of the gamble G….”
Mike,
I am still not happy with the might/would thing.
You think Oliver could say,
“It’s possible that you do not take the pill and I’m alive anyway.”
But, as you note, unless this possibility has a positive probability, it won’t help. But Oliver knows that his father took the pill, so the probability that his father did not take the pill is zero. So the probability of the conjunction is surely zero, too.
We need some counterfactual would/might conditionals to state what is relevant, it seems to me, and we are stymied by the conditional contradiction principle.
I am also unsatisfied by the evaluation of the lottery. You say,
But it depends also on the value of “I do not live”. And this is the part I worry about.
It would be quite interesting (and important) if there were an independent way to value the uncertain prospect; that could then be used to find the value of “I do not live”. But unfortunately that doesn’t seem to be in the cards.
But Oliver knows that his father took the pill, so the probability that his father did not take the pill is zero.
I don’t follow that. Oliver is asking about what his father might have done. He might have failed to take the pill and he (Oliver) might have lived anyway. He prefers the gamble in which this outcome occurs.
It is a little beside the point I guess, but I’m pretty sure most Bayesians will say that only propositions that are impossible get assigned zero. The door is always left open slightly for contingent propositions.
But it depends also on the value of “I do not live”. And this is the part I worry about.
Jamie,
What you risk in taking the gamble is what you now have. How good is that? Suppose the value of that is V. Now ask: is V greater than a .5 chance at V’, where V’ is existing in a better way? Non-existence has no value at all: it is neither good nor bad for you, so it neither adds nor subtracts from .5 x V’.
Is V greater than a .5 chance at V’…? I don’t understand that question.
Are V and V’ names for a state of affairs, or are they measures? I will assume they are states of affairs. So the question is whether V is better (for you) than a .5 chance of V’. But what is a .5 chance of V’? What will happen if that .5 chance doesn’t come through?
If you ask me, which is better: $5 or a .5 chance at $10? Then I understand. If I take the chance and don’t win the $10, then I just end up with whatever money I have now.
But that is the question of which is better, $5 or [.5 $10, .5 $0].
Do you see the problem? I can’t get a number for V’ or V unless I can already do the better-than comparisons. When I do those comparisons I learn how to place the states on a scale. This means, as far as I can see, that there is no way to put my non-existence on the scale. Of course, I could arbitrarily decide it should be at zero (seems like a nice spot for it), but unless I can compare it with other states I won’t be able to put the other states on the scale.