A couple of months ago, on Orangephilosophy, I posted descriptions of the following two lives:
Baby. A three-week-old baby, Baby, dies in an accident. Had Baby not died then, he would have enjoyed a happy childhood and adolescence, gone to college, entered a PhD program in philosophy, become a professional philosopher, and lived an enjoyable life until dying at age 80.
Student. A 23-year-old philosophy graduate student, Student, dies in an accident after a happy childhood and adolescence. Had Student not died then, he would have become a professional philosopher and lived an enjoyable life until dying at age 80.
I asked for opinions about whose death was worse for him (and therefore who I should murder, but that was really beside the point). Lots of people responded, and the responses were all over the map. Many people thought Student’s death was worse; many thought Baby’s death was worse; some thought they were equally bad. I think the view that Baby’s death is worse is the only plausible view, and at the risk of boring those who think this is obvious, I have some arguments.
First, take the view that the two deaths are equally bad. The only justification for this view is an Epicurean one: death is never bad for anyone, so both deaths are equally bad. People suggested a couple of reasons to think death isn’t bad. First, some say that death isn’t bad because the dead don’t suffer, so death doesn’t cause anything bad to happen. This is just to ignore the goods death prevents its victim from getting. Second, some say that you can’t harm a dead person – only the living can be harmed. But this is beside the point. The question wasn’t whether Baby or Student would be harmed more after death – the question was whose death is a greater harm. Obviously there’s more that can be said about Epicurus, but I find it hard to believe that anyone really believes the Epicurean view.
Several people said that Student’s death is worse. Some may have thought this had something to do with personal identity – Baby is so young that he is not identical with the (counterfactual) future person who goes to college and such. I think this is false, but if it matters, just substitute a two-year-old or some other suitably young person so that the person who dies is clearly identical to the (counterfactual) future person who does all the good stuff.
The more common reason offered in favor of Student’s death being worse was the fact that Student’s death frustrates many of Student’s actual, deeply held desires and projects; this is not true of Baby’s death. Having a desire frustrated is bad for a person; death is bad in this way for Student, but not for Baby; so Student’s death is worse.
But this is to make the same mistake the Epicureans make: it ignores another way in which Baby’s death is worse. Presumably, if desire frustrations are intrinsically bad, desire satisfactions are intrinsically good. (If not, tell me what is intrinsically good, and see if it makes a difference.) Baby is deprived of 23 more years of desire satisfactions than Student. So death takes 23 extra years of good stuff from Baby compared to Student. It also causes some extra bad stuff for Student, but there’s no way that could be sufficient to outweigh those 23 years of goodness lost. (Of course, this is all given the fiction that some desire satisfaction theory of well-being is true.)
The argument that Baby’s death is worse is straightforward. It starts with the claim that the overall value of an event for a person equals the intrinsic value of his actual life for him minus the intrinsic value his life would have had for him if that event hadn’t occurred. Call this claim C. (Feldman and Broome are among those who hold C or something like it.) C provides the most straightforward way to account not only for bad things caused by an event (like desire frustrations), but good things prevented by it (like desire satisfactions). Applying C to the cases, we first notice that had Baby and Student not died when they did, their lives would have been identical in value. So the worse death is suffered by the one whose actual life is less valuable. And it seems pretty obvious that Baby’s life is less valuable than Student’s, no matter what it is you think makes life worth living.
If you think Student’s death is worse, you have two options: deny C, or say that Baby’s actual life is more valuable than Student’s. If you deny C, you have to put something else in its place. (Suggestions welcome.) That’s what Jeff McMahan tries to do in The Ethics of Killing. In a later installment I’ll talk about his view. If anyone can produce a plausible view that entails that Baby’s life is better than Student’s, I would like to hear it.
Ben: I have a more detailed worry about personal identity which I’ll try and post later, but for now, let me ask a quickie about the implications of this view. On the account you’re advancing, wouldn’t it be the case that Baby’s death wouldn’t be as bad as, say, Fetus’s death (a 20 week-old-fetus, say), and Fetus’s death wouldn’t be as bad as, say, Embryo’s death? And, in the limiting case, Embryo’s death wouldn’t be as bad as The-Conceptus-That-Would-Have-Occurred-Absent-Successful-Contraception’s death?
Good question David. I think that, given that not much good happens to someone before birth, and certainly nothing good happens to them before consciousness begins, in many of the cases you list the deaths will be equally bad (not sure the last hyphenated one is actually a case of death though). That is, as long as the being that dies is the same one that would otherwise have received the good life, which I think is plausible.
I guess you might worry about this, thinking that the death of an embryo isn’t all that bad. I’m inclined to think that the death of an embryo is very bad indeed *for the embryo*. That doesn’t mean that it’s bad simpliciter, or bad for the universe. Maybe embryos lack some feature, like positive desert, or moral status, that an individual must have in order for an event that’s bad for that individual to be bad simpliciter. But of course this issue is pretty complicated, and I’m interested to hear what you have to say about identity.
Ben,
You assume, “had Baby and Student not died when they did, their lives would have been identical in value.” This is not necessarily true — at least, it is not necessarily true given your description of the two cases. Consider Velleman’s view: he thinks that the welfare value of a life is not equivalent to the amount of momentary welfare that that life contains. In particular, he thinks that misfortunes that lead to later successes don’t detract from the welfare value of a life as much as otherwise identical misfortunes that are suffered in vain do. (I take it that he’s appealing to the principle of organic unities here.)
Let’s assume for the moment that Baby and Student didn’t die when they did and that they both lived to eighty. Furthermore, let’s imagine that, although both Student and Baby suffered some misfortunes early in life, Student’s misfortunes lead to his later success and happiness, whereas Baby’s later success and happiness was just a windfall and Baby’s early misfortunes were just suffered in vain. (I take it that this is consistent with your description of the two lives.) Now if Velleman’s view is correct, then it’s false to say, as you do, that their lives are, ex hypothesi, identical in value; in actuality, Student’s life is of greater value than Baby’s life.
Therefore, it seems to me that you need to do one of three things: (1) refute Vellemen’s view, (2) redescribe your cases so as to rule out the possibility that the relationships between earlier and later moments in the two lives might be different, or (3) admit that your argument for the claim that Baby’s death is worse relies on certain assumptions about what makes a life worth living.
Doug,
I choose (2). Just stipulate that the lives would have been identical wrt Velleman-type values. Is there a reason I can’t do that?
Ben: I wasn’t sure, because I’m not familiar with McMahan’s arguments against C.
McMahan’s main arguments against C are sort of along the lines David was just suggesting, having to do with harms to embryos and such. Actually, the Velleman-type views are more of a problem for McMahan’s view than for C. Or so I say. C concerns the values of lives taken as a whole, so it has no problem with shape-of-life considerations.
Let me suggest an argument for the conclusion that Student’s death is worse.
Suppose we believe that what is good or bad for a person depends only on her actual desires, so that merely possible desires count for nothing. Call this view Actualism. On its face, Actualism seems quite plausible. To be sure, I might have desired to read continental philosophy; this is a possible desire of mine. But it’s merely possible. I do not actually desire to read continental philosophy, and so it would not benefit me to do so.
Now, consider your example of Baby and Student. It will be useful to name some possible worlds. Let W be the actual world. In W, Baby lives for three weeks and and Student lives for 23 years. Let X be the nearest possible world at which Baby doesn’t die when he actually did, and let Y be the nearest possible world where Student doesn’t die when he actually did. By your stipulation, in X, Baby lives to be 80, becomes a professional philosopher, and so on; and the same is true of Student in Y.
According to your claim C, the value of Baby’s death for him is given by the difference between the value of W for him and the value of X for him (the same being true of Student, replacing Y for X). And, according to Actualism, the difference in value for Baby between W and X depends only on the extent to which his actual desires — i.e. those desires he has in W — are satisfied in the two worlds. Presumably, Baby does have some actual desires, and some of these are better satisfied in X than in W. For example, Baby might actually desire not to be in pain; yet her dying in W might be painful. However, it’s plausible that the difference in Baby’s actual desire satisfaction between X and W is considerably less than the difference in Student’s actual desire satisfaction between Y and W. The key here is that some of Student’s actual desires are likely to be future regarding. For example, Student might actually desire, at the age of twenty, that he becomes a professional philosopher when he’s older; yet this desire is frustrated in W and satisfied in Y. By contrast, Baby can have no such future-regarding desires in W. He’s too young to even have the concept of the future. Of course, Baby may have some future-regarding desires in X. But these are merely possible desires, and, hence, excluded by Actualism. Plausibly, then, Actualism implies that Student’s death is worse than Baby’s.
Campbell – interesting idea. If I understand you right, you’re suggesting that even though all the same things happen to Baby in X and Student in Y, Baby’s life in X is not as good for Baby as Student’s life in Y is for student. So this is actually another possibility that I hadn’t considered before.
I don’t think what you’re calling Actualism could be right. The fact that you don’t desire to read continental philosophy shows (let’s assume) that you wouldn’t benefit from doing so, *given that you don’t have the desire*. But if you had the desire, and then you did the reading, *that* would be good for you, wouldn’t it? That’s what happens in the case of Baby. It’s not that in X, all these things happen to Baby but they fail to satisfy any of his desires. Baby acquires desires in X and then satisfies them, and that has to be good according to desire satisfactionism.
Suppose I could have decided to become a stockbroker instead of a philosopher, and that I would have satisfied just as many desires either way, but they would have been very different desires. Actualism entails that it’s really good for me that I didn’t become a stockbroker, because I would have had a really crummy life despite all my desire satisfactions. I would have been failing to satisfy lots of my actual philosophy-related desires, which is bad, and satisfying lots of non-actual stock-market-related desires, which counts for nothing. Clearly when evaluating someone’s life at a world we have to look at what desires the person has *at that world* if we’re desire satisfaction theorists.
Ben: You wrote: “I think that, given that not much good happens to someone before birth, and certainly nothing good happens to them before consciousness begins, in many of the cases you list the deaths will be equally bad …. That is, as long as the being that dies is the same one that would otherwise have received the good life, which I think is plausible.” This is only one way in which the analysis you give depends heavily on considerations of personal identity.
Consider the following alternate case: Student 1 would’ve lived the life you describe. Student 2 would have undergone fission the day after he died, in which case 2 people would’ve gone on to live qualitatively identical lives (of the sort you ascribe you Student).
Because of the logic of identity, Student 2 would not have survived fission (2 people would’ve resulted, and 2 does not equal one, so neither could be identical to the original Student 2). Fission would thus have “killed” Student 2. On your analysis, Student 1’s death is worse than Student 2’s, given that Student 2 wouldn’t have lived on much longer anyway (he was about to be “fissed”). But this seems wrong to me. If anything, Student 2’s death is *worse* for him, despite the fact he wouldn’t have survived fission. What this must mean is that it’s not the sameness of personhood that matters in assessment of value of (counterfactual) lives; rather, what matters are the psychological relations that are discontinued. In Student 2’s case, there would have been double those connections, each of which was discontinued by his death.
But then *this* suggests that if psychological relations — and not mere identity — are crucial, then Student’s death might be worse than Baby’s, simply because of the robust set of such relations that Student has — and Baby doesn’t — that are discontinued at his death.
(This is somewhat similar to worries about identity that attach to Marquis’ “Future-Like-Ours” account, which is itself quite similar to your account here.)
David – that’s an interesting case. I have lots of questions about what you say. Assuming I understand the example correctly correctly, I don’t share your view that Student 2’s death is worse, but I admit my intuitions are very shaky concerning such cases. Are you then committed to saying that if Student 2 underwent fission, and one of the post-fission people were immediately destroyed, leaving one person who is just like Student 2 would have been had no fission occurred, the whole situation would be just as good for Student 2 as it would have been had there been no fission? (There are the same psychological connections between Student 2 and post-fission person as there would have been between Student 2 at the earlier and later times if no fission occurred.)
You say that “it’s not the sameness of personhood that matters in assessment of value of (counterfactual) lives; rather, what matters are the psychological relations that are discontinued.” I don’t understand exactly what part of my view you think this poses a problem for, but I suspect it’s the idea that *lives* are the appropriate things to compare (since Student 2’s life ends at the fission.) My question is, if not lives, then what?
Yes, that’s right. But that’s also consistent with Actualism, as I’m thinking of it. If I were to desire to read continental philosophy, then that would be an actual desire, and so its satisfaction would be good for me. But I do not, in fact, have such a desire; hence, reading continental philosophy benefits me not at all.
Something similar may be said regarding the case of Baby. If Baby were to desire to become a philosopher, then his failing to become one would be bad for him. But he does not, in fact, have such a desire. To be sure, if had not died when he did, he would have desired to become a philosopher. That is, he does have that desire in X. But X is merely possible; it does not represent the world as it actually is. Baby’s desire, in X, to become a philosopher is a merely possible desire; hence, it’s irrelevant to the assessment of what is good or bad for Baby.
Campbell, I had thought you were arguing that Baby’s life in X is less valuable than Student’s life in Y. But you say: “If Baby were to desire to become a philosopher, then his failing to become one would be bad for him.” Then you agree also that if he were to desire to become a philosopher, and then became one, that would be good for him. So you must agree also that Baby’s counterfactual life would have been good for him were he to have lived it. So I don’t understand your objection anymore. Are you rejecting C, or do you think Baby’s actual life is no worse than Student’s?
It’s no part of my argument that Baby’s failure to get the things he would have desired, but actually doesn’t desire, is *intrinsically bad* for him. Baby has few desire frustrations. The badness of his death comes not from desire frustrations, but *prevention of desire acquisition and satisfaction*.
You say: “But X is merely possible; it does not represent the world as it actually is.” It is part of my view that things that happen at other possible worlds help determine the values of things that happen at this world. How else can you determine the values of events that prevent goods, or prevent evils? The things prevented are not actual.
Ben, perhaps I haven’t explained what I mean by Actualism very well. So let me have another go.
Actualism implies that:
(1) It is not the case that becoming a philosopher would be good for Baby.
The reason is that Baby does not desire to become a philosopher. She might have had such a desire; but, in fact, she doesn’t.
However, Actualism does not say that (1) is necessarily true; it allows that it might have been the case that becoming a philosopher would be good for Baby. (Note that this latter claim is consistent with (1)). Although (1) is true at the actual world, there are other possible worlds at which (1) is false. One such possible world is X. Thus, Actualism also says that:
(2) If Baby had not died when he did, then becoming a philosopher would be good for Baby.
But, of course, Baby did, in fact, die when he did. Hence, (2) is consistent with (1).
The following example may help to illustrate what I have in mind. Consider Beth the Baker. Beth bakes ten pies, but sells only five of them. Beth is disappointed by her low sales, and so she seeks our advice. She asks: “How might I have sold all my pies?” Suppose we reply: “That’s easy. You should have baked only five pies. Then you would have sold them all.” I take it that Beth would not be pleased by such advice. She already knows how to sell five pies; what she wants to know is how to sell ten pies. Our mistake is failing to realise that when Beth says “my pies”, this phrase acts as a rigid designator: at every possible world “my pies” picks out the pies that Beth bakes at the actual world.
Now, let W be the nearest possible world at which Beth bakes only five pies. And suppose that, at W, Beth sells all five pies. This might seem to imply that:
(3) If Beth had baked only five pies, then she would have sold all of her pies.
However, if we read “her pies” as a rigid designator, then (3) is false (but not necessarily false, because it’s true at W). In a similar way, Actualism says that, in the statement “Baby’s good consists in the satisfaction of her desires”, we should read “her desires” as a rigid designator. And that is why Actualism implies (1). On such a reading, it is not the case that, if Baby had not died when she did, many more of her desires would have been satisfied.
I hope that helps to clear things up. I should say that I’m not particularly wedded to Actualism. But it does seem a consistent view worthy of further consideration. In particular, it seems to capture what some people have in mind when they argue that Student’s death is worse.
Campbell,
Does Actualism say (1) that what is good or bad for a person at a time t depends only on what her actual desires are at t or (2) that what is good or bad for a person for now and for the future depends only on what her actual desires are at this moment? (1) seems consistent with what Ben is saying. (2) seems implausible. (2) implies that I should enter into Ulysses contracts that will ensure that my current desires about the some future time t will be satisfied even if my desires change so as to conflict with what I had previously wanted.
Ben: The fission case reveals something crucial about our assessment of value when applied to persons. Were Student 2 to be “fissed”, he would, by the logic of identity, “die” at that point. But that fact does nothing to undermine our assessment of the badness to him if he were to be killed the day before fission. His death would thus be just as bad (and, I think, worse) than Student 1’s death (your original Student case). This suggests that our assessment isn’t based on a counterfactual analysis of how *his* life would have gone, but rather on the value of the psychological experiences that would have occurred and would have stood in the right relationship to his own. The assessment, then, doesn’t go to “lives” (entities identical across time) but rather what I call “selves,” entities unified by strong psychological connectedness (I actually follow Parfit’s terminology here). But if it’s psychological relations that matter, and Baby has none of any significance, then Student’s death is worse than Baby’s.
(I’ve been meaning to post something on this argument for a while now, but haven’t yet gotten around to it. So I guess I’ve done so now!)
Doug,
Interesting question. I guess Actualism is neutral as between your (1) and (2). That is, (1) represents one kind of Actualist view, and (2) represents another. I’m not entirely clear on the contrast between (1) and (2), but my own inclination is to prefer (1). However, I don’t think that (1) is consistent with what Ben is saying, as you suggest. Notice that there is no time at which Baby (actually) desires to become a philosopher. Yet, presumably, Ben would say that there is some time t such that what is good or bad for Baby at t depends on whether or not she becomes philosopher.
Consider the respective degrees of social embeddedness. Baby has not forged nearly the rich web of relations that Student has. Student therefore will know that his death will cause a much larger number of friends, family and colleagues to suffer significant anguish. Ceteris paribus, this seems to make Student’s death worse for him.
Of course, how to quantify that as against Baby’s net comparative loss of desire satisfactions is unclear. Still, it seems clear that C isn’t the correct principle when you consider the case of Zygote and Baby. (They share similar biographical paths as in the example of Baby and Student, except that Zygote’s existence hasn’t been detected.) To be sure, Zygote’s life is poorer in desire satisfactions than Baby’s. But is it really therefore better for Baby to be killed rather than Zygote? Only an antichoice scruple would seem to motivate the answer yes.
(If you did answer “yes,” ask yourself at which point you’d rather die–at which stage you’d deem death less bad for you, given your assessment of all the morally relevant effects of your existential career.)
Campbell,
You say, “notice that there is no time at which Baby (actually) desires to become a philosopher.” Right. Indeed, that’s why Baby’s life is of much less value than Student’s life. But note that, had Baby not died when he did, there would have been a time when counterfactual-Baby actually did desire to become a philosopher and also a time when that actual desire of his would have been satisfied. Thus when we assess the value of counterfactual-Baby’s life, we conclude that its value is equivalent to the value of counterfactual-Student’s life. Of course, actual-Baby’s life is of must less value than actual-Student’s life given that Baby doesn’t have any actual-desire satisfaction, whereas actual-Student does have some actual-desire satisfaction. So, applying C, we conclude that Baby’s death is worse than Student’s death, just as Ben says.
Doug, you say:
But Baby did die when he did. Hence, there is no time when Baby (or counterfactual-Baby) actually desires to become a philosopher.
You continue:
If we evaluate counterfactual-Baby’s life according to counterfactual-Baby’s desires, then we allow that what is good or bad for Baby depends on counterfactual — hence, non-actual — desires. But that is precluded by your claim (1), which says that only actual desires count towards Baby’s good. To put things in your terms: Actualism requires that we evaluate counterfactual-Baby’s life according to actual-Baby’s desires.
Campbell – then are you saying, after all, that Baby’s counterfactual life *is* worse than Student’s counterfactual life? Because that’s what your version of Actualism seems to be committed to. Seems like a reductio of Actualism to me.
David – I think your view might be very similar to McMahan’s (who also follows Parfit closely in some ways), though I can’t tell until I know more about what you want to substitute for C. I’ll post my objections to McMahan’s view soon, and you can tell me if your view is different from McMahan’s.
Strange doctrines – the extra pain you say Student suffers from knowing how his family will feel is not the result of his death, but of his belief that he will die. Also, I never said it would be better for Baby to be killed than for Zygote, because I don’t buy the desire satisfaction theory – I was accepting it for the sake of argument. And when I ask myself whether I’d rather have died as a zygote or as a baby, I think it wouldn’t have mattered much, at least as far as my own well-being goes. But rest assured that none of my arguments are based on any sort of “anti-choice scruples.” I find C plausible, and I still haven’t heard an alternative to it.
Ben, you ask:
No. I claim only this: the difference in value between Baby’s counterfactual life and Baby’s actual life is less than the difference in value between Student’s counterfactual life and Student’s actual life. I reason as follows. Baby’s desires are equally satisfied in his counterfactual and his actual life. (Remember that I’m using “Baby’s desires” as a rigid designator, so that, in every world, it picks out the desires that Baby has in the actual world.) But Student’s desires are more satisfied in his counterfactual life than in his actual life. Hence, the difference in value is greater in Student’s case than in Baby’s.
As to the question of whether Baby’s counterfactual life is worse than Student’s, I find it hard to say. Here, on the Actualist approach, we are comparing the counterfactual satisfaction of Baby’s actual desires with the counterfactual satisfaction of Student’s actual desires. Since the desires are different in each case, the comparison is tricky. Fortunately, however, we need not make such a comparison for present purposes. The claim I made above, comparing the relevant differences in value, is entirely sufficient, given your claim C, to show that Student’s death is worse than Baby’s.
But those differences aren’t brute facts. They are grounded in the intrinsic values of the lives. So you are committed to saying that Baby’s counterfactual life is worse than Student’s, or that Student’s actual life is worse than Baby’s. Unless you’re using a kind of subtraction that I’m not familiar with.
When you say “Baby’s desires are equally satisfied in his counterfactual and his actual life,” it suggests that you think his actual life and his counterfactual life are equally good for him; since his actual life isn’t very good for him at all, you must think that his counterfactual life wouldn’t be very good for him either (despite all the desire satisfactions he’d have there). So it seems to me you’re committed to saying that Baby’s counterfactual life is worse than Students.
Campbell: I think that I am finally coming to understand Actualism, but let’s make sure, because now that I understand it, it seems extremely implausible. First, let me note that my actual wife is Erin and that my actual child is Fiona. Now let’s consider whether I would have been better off had I married someone else, say, Jane — in which case, I would have fathered Derek, not Fiona. On Actualism, we assess whether I would have been better off had I married Jane by determining how well my desires in this the actual world (e.g., my desire that Erin and Fiona be happy) would have been satisfied in this counterfactual world in which I marry Jane and father Derek. Now I’m a fantastic husband (so I say), so let’s assume that Erin would have been considerably less happy had she married someone else and that Jane would have been considerably happier had I married her. And, of course, Fiona wouldn’t be happy at all had I married Jane; Indeed, Fiona would never have existed had I married Jane. But we can suppose that Derek would have been very happy had he existed. So, on Actualism, we should conclude that I would have been much worse off had I married Jane even though, had I married Jane, I would have desired that Jane and Derek be happy and those desires would have been satisfied. Isn’t this an implausible way of assessing whether I would be better off in some counterfactual situation? Shouldn’t we assess whether I would have been better off had I married Jane by looking at what I would have desired had I married Jane and looking at whether I would have had more counterfactual-desire satisfaction in this counterfactual world than I have actual-desire satisfaction in this actual world?
Doug, you ask:
My answer: I think that remains to be seen. As I said to Ben, I’m not strongly in favour of Actualism, but I do think it’s an interesting view worthy of further investigation. I must admit that I was quite resistant to Actualism when I first learnt of it. (I learnt of it from my friend Josh Parsons, who has a good paper on the topic: Axiological Actualism, Australasian Journal of Philosophy, 80 (2002), pp. 137–47. Apologies to Josh if I’ve mangled his view!) But now it’s kind of grown on me. One reason that I’m interested in it is that it may help, or so I suspect, in the notoriously tricky area of variable population ethics. Another advantage of Actualism is that it may explain the intuition that many people have regarding Ben’s example — i.e., that Student’s death is worse. But, in my opinion, we don’t yet know enough about the advantages and disadvantages of Actualism to decide just how plausible it is.
Campbell – the position you seem to be endorsing is not Actualism in Parsons’ sense. Parsons describes Actualism as the view that we should not assign levels of welfare to merely possible people. There are no merely possible people involved in this example.
Ben, that’s right: the view I’ve tried to articulate is not the view that Parsons defends. But I think it is a natural extension of Parson’s view as applied to the particular case of desire satisfaction theories of wellbeing. Think of it this way. Merely possible people do have merely possible desires. Thus, if we want to hold a desire satisfaction theory of wellbeing, but we want to resist assigning levels of wellbeing to merely possible people, then a natural thing to say is that merely possible desires do not count toward wellbeing. But I wouldn’t want to say that Actualism, as I’ve described it, is implied by Parsons’ view.
More generally, we might say the following. Metaphysical Actualism is the view that mere possibilia do not exist. And Ethical Actualism is the view that mere possibilia have no moral significance. (It might be thought that Metaphysical Actualism supports Ethical Actualism. How can we attach moral significance to something that doesn’t exist?) Applying Ethical Actualism to desire satisfaction theories of wellbeing yields the view that merely possible desires do not count toward wellbeing (otherwise a certain kind of mere possibilia — i.e. merely possible desires — would have moral significance).
Anyway, that’s how I was thinking of things.
Campbell – Now it sounds like what you really want to do is reject C. C determines the value of an event for a person by appealing to the value of a life the person would have lived had the event not occurred. But, on the view you’re now describing, there is no such life, since it’s not actual, so there’s no second term for comparison. (Note that if that’s the view, you can’t talk about “differences” in the way you have been.) So the version of Actualism you’re putting forth here is very radical – it evidently entails that facts about what *would have* happened are never morally relevant. Again, I take that to be a reductio of the view. If I cause you not to be happy, I can’t justify my behavior by pointing out that your happiness is merely possible, and therefore not morally relevant.
Ben, I think that C is consistent with Actualism. You raise a difficult issue, and I’m not sure I can deal with it adequately. But I’ll have a go (I’m getting out of my depth as a metaphysician here).
As I understand it, your argument is as follows. C implies that we can compare actual lives with merely possible lives. But, according to Actualism, merely possible lives do not exist; in which case they cannot be compared with anything. Hence, C is inconsistent with Actualism.
Here’s the Actualist response. Actualism does not deny that merely possible lives exist; rather, it denies that they are lives. I gather that Actualists diverge on the question of what mere possibilia are, but the following is one proposal (it’s the one that Parsons gives in the paper I cited earlier). Mere possibilia are representations. In the present case, a merely possible life is a representation of a life. Such representations are themselves actual, and so there existence is perfectly kosher by Actualist lights.
Consider again the example of Baby. When we ask how greatly Baby was harmed by his death, C tells us to compare Baby’s actual life with the life he would have had if he had not died when he did. On the current proposal, the latter life, being a merely possible life, is understood as a representation of a life: specifically, it represents what Baby’s life would have been like if he had not died when he did. Thus, in order to apply C, we ask whether this merely possible life represents a life that is better for Baby than Baby’s actual life.
Now, on a desire satisfaction theory of wellbeing, the comparison just described is made according to Baby’s desires. We ask whether the merely possible life represents a life in which Baby’s desires are more satisfied than they are in his actual life. But, given Actualism, we must consider only Baby’s actual desires. The reason is that nonactual — i.e. merely possible desires — are not desires at all. Rather, they are representations of desires.
Anyway, that’s enough amateur metaphysics from me.
Campbell – but the desire *satisfactions* in the merely possible life are also just representations of desire satisfactions. So they aren’t really desire satisfactions either, right? In which case no merely possible life has any value for anyone.
The talk about possible lives being mere representations is a red herring. If I describe a life I might have had as a life where I satisfy many desires I would have had (but actually don’t have), haven’t I just represented that life as being really good for me?
On reflection, I feel that some of what I’ve said above is confused and misleading. (I hope you’ll forgive this; I’m making it up as I go along, after all.) So let me now try to clear up two misconceptions in my earlier comments.
1. I may have given the impression that the view I was aiming to articulate was better supported by plausible metaphysics than the opposing view. In particular, I may have suggested that the opposing view, which deals in merely possible desires, is guilty of attaching moral significance to metaphysically dubious entities. I now think that was a mistake. It’s difficult to do much philosophy at all without modal talk — i.e. possible worlds and the rest — and ethics is surely no exception. On its face, such modal talk presents a problem for metaphysical actualists. Statements about merely possible worlds seem to carry non-actualist ontological commitments. But actualists have various strategies for analysing modal talk in an actualist-friendly way. I described one such strategy in my previous comment: talk of merely possible things is understood as talk of actual representations of things. This being the case, metaphysical actualism should have no quarrel with a theory of wellbeing that deals in merely possible desires. Actualists may simply employ their favourite actualist analysis of modal talk to explain away any dubious ontological commitments that talk of merely possible desires might suggest. I should also concede that the view I earlier called Ethical Actualism — i.e. the view that mere possibilia have no moral significance — is highly implausible, as I think you rightly pointed out.
2. I may have given the impression that the view I was trying to articulate followed simply from the definition of a desire satisfaction theory of wellbeing. The opposing view considers, not only the satisfaction of desires, but also the satisfaction of merely possible desires, which are not really desires at all, but mere representations of desires (or some such thing). And this might be thought to suggest that the opposing view isn’t really a desire satisfaction view at all. But I’m now inclined to think that this is mistaken too. After all, merely possible desires are representations of desires, and this should be sufficient to qualify the opposing view as a desire satisfaction view. I think the best thing to say here is simply that there are two interpretations of the desire satisfaction theory of wellbeing: one is represented by the view I called Actualism, and the other by the opposing view. Deciding between these two interpretations will need to proceed in the ordinary way: by asking which view better coheres with our considered intuitions, and so on. Questions about the ontological status of merely possible desires are not at issue here.
Now, having said all that, let me have a go at describing the two competing interpretations just mentioned. Consider again the example of Baby (leave Student out of it for now). And consider, in particular, this question: was baby harmed by his death? (This is a little simpler than the question posed in the original example, which asked, not only whether baby was harmed, but what the extent of the harm was.) As before, let X be the nearest possible world at which Baby does not die when he did. On our actualist analysis, X is a representation of a way the world might have been (though not a way that it is): it represents Baby as becoming a philosopher, living to the age of 80, and so on.
Your claim C, as I understand it, implies that Baby was harmed by his death just in case his level of wellbeing in X is greater than his actual level of wellbeing (where wellbeing here is a measure of the goodness of a person’s whole life). Thus, if we accept C, we need to ask whether X represents baby as having a higher level of wellbeing than he actually has.
On a desire satisfaction theory of wellbeing, Baby’s (actual) level of wellbeing is given by the extent to which the (actual) world is as Baby desires it to be. That part is clear enough. But what is Baby’s level of wellbeing in X? (Or, more precisely, what level of wellbeing does X represent Baby as having?)
Here there are two possible views:
(DS1) Baby’s level of wellbeing in X is given by the extent to which X represents the world as Baby desires it to be.
(DS2) Baby’s level of wellbeing in X is given by the extent to which X represents the world as X represents Baby as desiring it to be.
DS1 is the view that I earlier called Actualism. It requires that we measure Baby’s wellbeing in X according to his actual desires. DS2 is the opposing view. It requires that we measure Baby’s wellbeing in X according to the desires that X represents Baby as having — i.e. the desires that Baby would have had if he had not died when he did. For reasons given in earlier comments, it’s likely that DS1 and DS2 will diverge on the question of whether Baby was harmed by his death. Assuming that Baby dies painlessly, DS1 implies that Baby’s level of wellbeing in X is equal to his actual level of wellbeing; whereas DS2 implies that Baby’s level of wellbeing in X is greater than his actual level of wellbeing. Hence, DS1 says that Baby is not harmed, while DS2 says that he is.
Which view is superior? Hard to say (I reckon). In this particular example I’m inclined to say that my intuitions are on the side of DS1. Painlessly killing a three week old baby causes no harm (to the Baby), just as painlessly killing a dumb non-human animal, or a human embryo causes no harm. Other people’s contributions to this discussion, and to the earlier discussion on Orange Philosophy, suggest that they share this intuition. But clearly not everyone does. My hunch is that the debate between DS1 and DS2 will not be resolved so easily.
Ben, I’ve been thinking some more about an objection to Actualism that you made earlier. And I now think that there’s more to be said in defence of Actualism. So, here it goes …
To focus on the particular objection I have in mind, let me describe a simpler example, which brings out the objection more clearly. Phylis is a philosopher; she has the typical philosopher’s desires (to write good papers, to impart knowledge to her students, etc.), which are pretty well satisfied. Lola is a lawyer; she has the typical lawyer’s desires (to make money, to deceive people, etc.), which also are pretty well satisfied. Suppose we hold a desire satisfaction theory of wellbeing that makes interpersonal comparisons of desire satisfaction in such a way that Phylis and Lola are considered to be equally well off; they have the same level of wellbeing.
Now, suppose that there was a pivotal moment in Phylis’s life when she decided to study philosophy rather than law. If she had chosen law instead of philosophy she would never have developed the philosophy-related desires that she has; rather, she would have come to have the typical lawyer’s desires, just like Lola. How well off would Phylis have been if she’d chosen law instead of philosophy? Or, to put the question another way: if X is nearest possible world at which Phylis chooses law, what is Phylis’s level of wellbeing at X? According to Actualism, we must judge Phylis’s wellbeing in X by the desires she has in the actual world — by the typical philosopher’s desires, that is to say. Plausibly, then, Actualism implies that Phylis’s level of wellbeing is lower in X than it actually is; if Phylis had chosen to become a lawyer, she would have been worse off.
This implication may seem puzzling, and one way in which to bring out the puzzle is by comparing Phylis’s wellbeing in X with Lola’s. Presumably, Phylis’s choosing law would not have changed Lola’s choice of profession (at any rate, we may suppose that this is so). Thus, in X, Lola’s life is just as it is in the actual world. Hence, Actualism will conclude that Lola is no worse off in X than she actually is; Phylis’s choosing law would have made no difference to Lola’s wellbeing. But now we are led by Actualism to the conclusion that, in X, Phylis is worse off than Lola. But that seems quite odd, especially if we suppose, as we may, that the desires that Phylis has in X are just as well satisfied in X as Lola’s are. Actualism conflicts with our intuition that Phylis and Lola are equally well off in X. As you suggested earlier, this kind of example might be regarded as a reductio of Actualism. But I think that would be too quick. There is something to be said in defence of actualism here.
Suppose that I had begun by introducing the example differently — in particular, by describing X as the actual world. In that case, we would have considered Phylis’s desires in X as her actual desires; in which case Actualism would have implied that Phylis and Lola were equally well off in X, just as our intuition tells us. This reveals a kind of relativism in Actualism: Phylis’s level of wellbeing in X depends on which world is considered as actual. When the world in which Phylis is a philosopher is considered as actual, Actualism tells us that Phylis is worse off than Lola in X; but when X is considered as actual, Actualism tells us that Phylis and Lola are equally well off in X. This kind of relativism may sound incoherent. But, fortunately for Actualism, two dimensional modal logic has the resources to make sense of this kind of thing. Now, I’m no logician; so the following may be a little crude. But I think it captures the basic idea. In 2D modal logic, attributions of truth values to sentences are doubly relativised: firstly, there is the world at which the sentence is said to be true or false; secondly, there is the world that is considered as actual. For example, consider the sentence:
(a) Phylis and Lola are equally well off
And let Y be the world in which Phylis is a philosopher (as described above). In the logic just described we can say that (a) is true at X with X considered as actual, but that (a) is false at X with Y considered as actual.
So, Actualism might be defended as follows. When we judge intuitively that Phylis and Lola are equally well off in X, we are thinking of X as though it were actual. Hence, our intuition supports only the claim that, when X is considered as actual, X represents Phylis and Lola as equally well off. But Actualism is perfectly consistent with that claim. Actualism is not consistent with the different claim that, when Y is considered as actual, X represents Phylis and Lola as equally well off. But the latter claim is not supported — or, at any rate, not as well supported — by intuition.
Campbell – I’m just not seeing any motivation for the desire satisfaction theorist to evaluate how well someone’s life goes for him at a world based on desires he doesn’t have at that world. And I just don’t see any reason, when evaluating Phylis and Lola’s lives at X, to think it matters whether we are “thinking of X as actual.” Your sentence (a) seems false (by desire-satisfactionist lights) no matter whether we are thinking of X as actual or not. I suppose one motivation is just to get a certain answer in the Baby/Student case. I’d regard that as a pretty weak reason.
All right, here’s an argument. Suppose that there’s a career-changing machine. If you were to choose to go through the machine all of your philosopher-related desires would be replaced by lawyer-related desires. Moreover, your life would be altered just so much that your new desires would be just as well satisfied as your old ones. Finally, suppose that you are offered five dollars to go through the machine and that you decline the offer. According, to the non-actualist desire satisfaction theory, you have harmed yourself by declining the offer. If you had gone through the machine the lawyer-desires you would have had would have been just as well satisfied as the philosophy-desires you actually have; and, in addition, your desire for money would be a little better satisfied, because you would have gained five dollars. According to the actualist desire satisfaction theory, on the other hand, you have not harmed yourself. I submit that the latter conclusion is more intuitive.
Campbell – that is an interesting case. I wonder if the intuition has to do with the fact that a machine is involved. I suspect that if the example were reversed – the machine keeps your desires exactly the same, whereas they would otherwise change “naturally” – intuition would still say not to go into the machine.
But my bigger concern is something else. I think this example brings out a bad feature of actualism, which is that it runs afoul of a plausible principle. If I remember correctly the principle is called the principle of normative invariance (I think Erik Carlson came up with it). According to PNI, the normative status of an alternative does not depend on whether the alternative is performed. In your desire machine case, actualism entails that I should not get into the machine — *but only given that I don’t actually get into it*. If I do get into it, then it would seem that actualism would entail that I do the right thing, and that not getting in would have been wrong. I think PNI is extremely plausible. If actualism is inconsistent with it, I consider that to be a big problem for actualism.
Ben, a couple of replies.
1. Regarding your suggested modification of the example so that ‘the machine keeps your desires exactly the same, whereas they would otherwise change “naturally”‘, it’s important not to conflate modal and temporal requirements. Actualism imposes a modal requirement: it says that only actual desires matter. A different kind of view — call it Presentism — imposes a temporal requirement: it says that only present desires matter. (I think Doug has defended a kind of Presentism in another PEA Soup thread.) Now, in the modified example, suppose that you do not in fact go into the machine. Then the new desires that you come to have naturally are (among) your actual desires. Your actual life is such that at every time t, the actual desires you hold at t are satisfied; whereas, the counterfactual life you would have had if you’d gone through the machine is such that at some times t, your actual desires at t are not satisfied. So Actualism again implies that you have not harmed yourself by declining the offer of going into the machine.
2. Regarding PNI, I agree that this does raise a serious problem for Actualism. But I’m not convinced that it’s insurmountable. Notice that Actualism is merely a theory of wellbeing; it purports to tell us what is good or bad for people, but it makes no claims about what people ought to do. In order to generate claims of the latter kind, Actualism needs to be combined with a view regarding the connection between wellbeing and right action. To keep things simple, suppose we go in for some kind of egoism, so that what a person ought to do depends solely on her own wellbeing. Then we have two options. Here’s one:
(P1) An action A is permissible iff the outcome of A is at least as good for the agent as every alternative.
(Here I’m thinking of outcomes as possible worlds.) Actualism implies that the truth of the right-hand side of this biconditional may depend on whether or not the action A is actually performed. Hence, the conjunction of P1 and Actualism implies that the permissibility of performing an action may depend on whether or not the agent performs it. And this would violate PNI.
But we have another option instead of P1:
(P2) An action A is permissible iff, on the hypothesis that A is actually performed, the outcome of A is at least as good for the agent as every alternative.
The idea is that, when judging the permissibility of an action A, we do so under the supposition that the outcome of A is the actual outcome. Of course, on a non-actualist theory of wellbeing, P1 and P2 come to the same thing. But, given Actualism, these two principles have importantly different implications. Suppose that the action A is not in fact performed. Then, given Actualism, the (actual) goodness of A’s outcome for the agent, may differ from the goodness that it has on the hypothesis that A is actually performed (since that hypothesis might alter which desires are regarded as actual). Now, the truth of the right-hand side of P2 is independent of whether the action is actually performed. Hence, the conjunction of Actualism and P2 is consistent with PNI.
That’s all rather quick and sketchy. But I hope it gives the basic idea. I believe that this kind of approach raises other difficulties, which I won’t go into here. My basic point is that Actualism is not, by itself, inconsistent with PNI.