Update (13 August, 2007). I’ve now written a very short paper on this issue: How to Live a Life Worth Living. As you’ll see, it was significantly influenced by useful comments I received here.
The recent discussion of McTaggart initiated by Kris has been very interesting. Among other things, it has got me thinking about the notion of “a life worth living”. Although ubiquitous in population ethics, this notion resists easy analysis. One wants to say that a life is worth living just in case it would be better to live it rather than live no life at all. But on reflection, that seems mysterious. How does one live no life at all? It seems like one of the relata of the “better than” relation has gone missing. We’re trying to compare something, a life, with nothing.
Here I shall propose an analysis that avoids such mysterious comparisons. On my proposal, whether a life is worth living depends solely on whether it is better than certain other lives.
Instead of giving an elaborate explanation of the analysis, I’m just going to state it very austerely. I hope that’ll be enough to get the discussion rolling.
Let L be a possible life, a life lived by someone in some possible world. Assume that L is composed of temporal parts, which obey classical mereology. In particular, for any parts of L, there is a part of L that is a fusion of those parts. Assume that some parts of L are simple, i.e. they have no parts. Call these slices of L. Assume that the slices of L are ordered by an “earlier than” relation.
Say that F is a future of L iff F is a proper part of L and no slice of F is earlier than any slice of L that is not a slice of F. Assume that for any future F of L there is a unique possible life L-F such that L-F is an intrinsic duplicate of the largest part of L that doesn’t overlap F, i.e. the mereological complement or remainder of F in L.
Now here’s the view:
- A future F of a life L is worth living iff L is better than L-F.
- A life L is worth living iff some future F of L is such that every future of which F is a part is worth living.
Interesting.
I suppose that lives have some extremely short parts at their beginnings. Call the very first extremely short part a ‘flash life’. (There may be no shortest, in which case just let the first nanosecond long life part be a ‘flash life’.)
Lives are worth living, according to your view, if they are better than their flash lives. Is that about right?
It may be very difficult to compare flash lives to lives of normal length. And it seems, intuitively, that all flash lives are equally good. Well, to me it does. So then maybe the value of a flash life itself fixes “barely worth living”.
It’s altogether possible I’m misunderstood the proposal completely. It seems to me, however, that it’s very difficult to capture the notion of a life worth living via a comparison between stages of life like this. As I’m reading the proposal, roughly stated, it’s that a “future” is worth living if it’s better than its mereological counterpart (i.e., the part of the life that is not part of the future, L-F; for the sake of simplicity I will call this “a past”). And a life is worth living so long as there is some future that is better than it’s counterpart, it’s past. But this seems to give both false positives and false negatives. Assume that the past (L-F) is just awful. I mean, hell on wheels. But then the future (F) is hell on wheels with one additional lollipop, say. So F is better than L-F. On this view, F is worth living. But why should this be the case? Why should a live of extreme torture with one lollipop, though better than a life of extreme torture with no lollipops, be worth living? Surely it is more worth living, but clearly not worth living, as it were, all things considered. Also assume that F is right at the end of the life, so that “every future of which F is a part” passes the test in (1). Again, this appears to deliver the verdict that the life is worth living.
There might also be false negatives. Assume that the past (L-F) is among the greatest possible lives that ever could be experienced. If you’re a hedonist, pure bliss. If you’re a perfectionist, I dunno, Leonardo or something. Leonardo + Einstein. Then assume that the future (F) is only slightly worse – pure bliss with one short hangnail. But then this F, on your view, would not be worth living. Again, this strikes me as wrong. Now you might say that there is still some F, such that every future of which F is a part is worth living (entailing the whole life is worth living, even though that F is not). But assume that the life has a steady downward trajectory from birth until death, but that its beginning is SO GOOD, and the losses SO SMALL that the quality of the life at the end is still something close to, say, pure bliss. Maybe you just want to deny that any life can have that sort of shape. But even so, the view seems to generate false positives (see above).
It strikes me that it’s difficult to get a notion of “a life worth living” without a substantive notion of value that might be picked out differently by different accounts of welfare (i.e., a “life of human dignity” or something like that). But it seems to me that it’s unlikely to be picked out by a comparative like “better-than,” because problems like this could always be run these sorts of relations, no? Maybe “better-than” could accommodate “more worth living,” but this, it seems to me, is just another way of stating “better-than”.
Or perhaps I’ve just misunderstood the proposal (which is perhaps the easiest explanation).
It’s an interesting puzzle but I worry that something has gone wrong in stating it. Compare:
1) “One wants to say that a life is worth living just in case it would be better to live it rather than live no life at all. But on reflection, that seems mysterious. How does one live no life at all? It seems like one of the relata of the “better than” relation has gone missing. We’re trying to compare something, a life, with nothing.”
with analogical
1*) “One wants to say that a beer is worth drinking just in case it would be better to drink it rather than drink no beer at all. But on reflection, that seems mysterious. How does one drink no beer at all? It seems like one of the relata of the “better than” relation has gone missing. We’re trying to compare something, a beer, with nothing.”
Now, 1* seems pretty funny. There doesn’t seem to be much mystery there. What would make 1 any more mysterious?
Dale, that’s not quite right. The future is worth living if the entire life is better with it than without it (where the life without it is just the life cut short). A worry is that still, a Leonardian life plus a somewhat more ordinary future tacked on might be worse than the Leonardian life alone, even though that future is well worth living. But Campbell could just say that his notion doesn’t capture exactly the intuitive concept expressed by the words “life worth living”, and maybe there is no single concept there anyway.
Jussi, the difference is that in the case of beer (so to speak), we can compare two things that are easy to understand: drinking beer and drinking no beer. We don’t have to compare a beer with its nonexistent counterpart, whatever that would mean. Whereas in the case of the life, it’s a lot harder to understand what it is to live no life at all. What, exactly, is that relatum?
James (if I may) –
Ah! I see. I misread “L is better than L-F” as “F is better than L-F”. But even so, it still seems to give the false positive. (I’m not quite sure I want to agree with your alternative counter-example; seems to me that that a Leonardo-type life with an ordinary future could still be better with the ordinary future than without it.) Surely a life of torture with one second of sugary happiness before death is better than that life cut short before the sugary goodness. But that surely doesn’t mean that the life is worth living.
Perhaps it’s the case that this proposal is not meant to capture the intuitive notion of a life worth living. But then it seems to me that calling it a “life worth living” is a bit misleading. It would perhaps be better to say the life has some other property (call it a “life worth x-ing”), interesting perhaps, that may or may not correspond to the forensic notion of a life worth living, the one that makes a difference to morality and moral obligation, i.e., as when we say “it’s unacceptable to let people live lives that are not worth living when you can do something to prevent it,” or something like that. The latter property could be captured by some more substantive notion, say, basic human dignity or something like that, that wouldn’t necessarily involve any mysterious comparisons.
Jamie,
I did think about that. But, I’m not sure I see the problem. I worry that the problem is created by the misleading surface-grammar of ‘to live no life’. This shouldn’t be thought of as for some x, I live (x) like ‘to drink no beer’ should not be thought of as for some x, I drink (x).
However, it does not seem to be too difficult to grasp the thought of not being alive, i.e., being dead. Of course there is nothing it would be like but that just is supposed to be the point. For what it’s worth if I had to say how to compare whether a life is worth of living, I guess the question should be posed from the first-person perspective. Would there be reason, given what is in store in the rest of the life, to go on living or would just calling it a day be more choiceworthy? That sounds like a sensible question and it would be the way I would think about whether life is worth living. Of course, I’m passing the buck here, but the missing relata problem doesn’t seem too bad.
Hi Dale.
Right. But Campbell’s view doesn’t say that that life is worth living. That life, I presume, does not have a future, F, such that every future of which F is a part, is worth living. It has a future that is worth living. But many futures that have that one as a part, are not worth living. Anyway, that’s the idea.
Jussi,
That’s perfectly fine for comparing two lives, one of which is just like the other but cut short. That’s not the problem, since there are, plainly, two perfectly comprehensible relata. The problem is comparing a life with non-existence. Can you do that from the first person perspective?
In an earlier post, I discussed a view of what lives are according to which one has a life even in a world in which one does not exist.
It seems to me that a view like Campbell’s could be developed with the same advantages. Instead of assuming that a life is an event with temporal parts, assume that a life is a *set* of events. Your life at w is the set of events that you participate in at w. (Campbell’s idea of L-F is could be construed as the set of events in L not in F.) Assume that, just as events can have value, sets of events have value. The empty set has no value. A world in which you don’t exist is a world in which the empty set is your life. If your actual life is going badly, then the value of L is less than the value of the empty set, which is the life you would have had, had you not been born.
If one insists that one needs relata to make comparative judgments about lives, this seems to do the trick.
Why not? It does require some metaphysically weird thought-experiments. Imagine you were teleported back in time to the ‘original position’ before your birth – to a time when you were a mere soul. At that point, you are given a choice: whether to get the described life or no life at all. Of course, there is nothing like the living that life would be like (as there is nothing like not living the rest of your life would be like). The question about the value of life then is a a question of choiceworthiness of the lives. Do the elements of the life give enough reasons to choose it or against choosing it?
I know the thought-experiment is metaphysically weird but it’s just a useful story to think about the issue.
I don’t think that’s a good test, Jussi.
First, there I am, in the original position. So, I exist, so I already have a life. So, it seems to me that I am choosing whether to augment my life by the one described, or not to augment it. That’s a different question.
Second, leaving the first objection aside, your test is a test of whether a life is choiceworthy. That is one way to interpret the question of whether it is worth living, but I think it’s not the one that Campbell and most people have in mind. Take one of the horrible lives, the ones full of torture — a life that is much worse for you, as we’d say if we weren’t worried about the metaphysics, than no life. But it could be enhanced by some features that are instrumentally good for millions of other people. This might make it in no way better for you, but make it highly choiceworthy.
Now some people will deny this — they’ll think that a life loaded with sacrifice is thereby very good for its bearer, maybe. But those of us who don’t buy that will not accept the test. In general terms: there are many features of lives that make us prefer to live them aside from what makes them better to live.
It’s not perfect but I think it’s workable. I’m not sure given a choice in the original position would constitute a life. It’s not that it would continue to be like something unless one chose the life. And, of course, it is a question we ask from that perspective hypothetically and we have our own lives that are not at issue anyway.
About the second point. I think my intuition *is* that life worth living is more to do with choiceworthiness of life than how good the life is for person whose it is. I think many lives are worth living even when they are miserable for the person whose lives they are just because they make a huge contribution to the world. Was van Gogh’s life worth living? It was bad for him but the paintings make so many lives so much better. My intuition is that his life was still worth living – even for him.
In any case, if you want to associate the worth of the living with goodness of the life, there is still a possibility to substantially limit the reasons that are taken into account from the ex ante perspective. You might think that the considerations which make lives better or worse for the agent are the relevant group of reasons. Are those reasons sufficient for choosing life over not being born?
I know all these are difficult substantial questions. But, I just don’t think the worthiness of a life is something philosophy is going to give a nice answer in terms of a definition. We are just looking at a way in which we could think about the matter in life.
These are the conditions.
1. A future F of a life L is worth living iff L is better than L-F.
2. A life L is worth living iff some future F of L is such that every [future] of which F is a part is worth living.
If I am reading these right (1) requires that future F of L is worth living iff. F is on balance good. (2) is more difficult. I think (2) requires that there is a possible future F of L such that every other possible future F’ (of L) that includes F is on balance good. I don’t know why that condition is imposed. Now I bracketed [future] above since I think you might have meant to put ‘life’ rather than ‘future’ there. I don’t know.
Here’s a question about (2) as it is written. Why must the future of L be worth living in order for L to be worth living? Suppose L-F is most of my life and extremely good. Suppose the future of F of L (say all possible futures F of L, if you like) are on balance just slightly bad. Still my life L might be on balance terrific. So why is L not worth living? Or rather, why is (2) a condition?
Jamie,
Aha. (Again.) Very good. I now see why neither of my cases are counter-examples.
One wonders, though, whether this account of a life worth living could be workable without a substantive conception of a life worth living such as the ones I’ve been proposing so far. Perhaps the following is what I’ve been fumbling toward. Say, for instance, we’re interested in whether L is better than L-F. On what grounds would we say this? Well, presumably different accounts of welfare will say different things. Possibilities:
Hedonism: whether F has at least +1 hedons.
Preference satisfaction: whether the agent would prefer (say, with perfect information, possibly with an experience of both, etc., etc.) L to L-F.
Perfectionism: whether F conforms to the minimum basic standard for human beings (say, “basic human dignity”).
The list goes on. In each of these cases the view offers a distinctive account of what makes a period of life better than the absence of that period of life. But then if this is the case, why can’t we simply rely on that to give us an account of a life worth living? If we have to have recourse to it anyway, why not simply say that a life is worth living when it’s overall balance is, e.g.,+1 hedons? When it conforms to basic human dignity? When it would be preferred over no life at all (perhaps this is the mysterious comparison that Campbell was worried about)? Of course, this would not mean that the above proposal is false, per se, but only that it seems to crucially rely on the answer to a question that might, itself, be an answer to the question of when a life is worth living. In other words, it might be that in order to figure out when, on this proposal, a life is worth living, we need to know when a life is worth living. (Maybe on the preference view we need something like a “flash life”.)
Lots of very helpful comments. Thanks to everyone, especially Jamie (James?), who seems to be doing my job for me.
A few replies.
Jamie. The thing about flashlives is basically right (though the view seems less clever when you put it that way). I agree, comparing a life with a flashlife is odd. But it’s not as odd as comparing a life with nothing, or with some kind of artificial empty “life”, e.g. the empty set. So this seems to me a more satisfactory solution to the Problem of the Missing Relatum.
Jussi. You say:
This is roughly how I propose to think about whether a future is worth living (except for the choiceworthy bit, which Jamie criticised already). I say a future is worth living iff the whole life including the future is better than the life with the future removed, i.e. it’s better that the life continues through to the end, rather than stopping immediately. But from the fact that a certain future is worth living we cannot infer that the life itself is worth living. Dale has given a good example of this. Imagine a life that is all torture except for the last minute, which is bliss. Once you get to the last minute, it’s worth living it, but it’s not worth living through the rest of the life to get there in the first place. The last minute is a future worth living (in the slightly artificial sense of “future” I’ve stipulated), but the life as a whole is not worth living.
So the question is: how do we get from futures worth living to lives worth living? In the case of futures, there’s no missing relatum problem. The two relata are (i) the life with the future and (ii) the life without the future. In the case of lives it’s not so simple. We can’t say the two relata are (i) the life with the life and (ii) the life without the life. If you remove the life from the life, you’re left with nothing; one relatum vanishes.
My proposal is a way to overcome this problem. I try to use something like the mathematical concept of converging to a limit. We keep making the future bigger and bigger (and hence the life without the future smaller and smaller) until we get convergence. As Jamie points out, this might make one relatum get very, very small, becoming a “flashlife”. But at least it never vanishes altogether.
Kris. The view you suggest is exactly the kind of thing I was trying to avoid, filling in the missing relatum with an artificially constructed “empty life”. I’d rather not say that any set is a life, let alone that the empty set is one. That’s bizarre; better to do without it if we can. My analysis is supposed to show how we can do without it.
Mike. I did mean to write “future” in my analysis. I suspect the problem might be that I’ve stipulated a slightly artificial definition of “future”. In my terminology, a future of a life is just any continuous chunk of the life that goes all the way to the end. So, while the end point of every future is fixed (it’s just the end of the life), the starting point varies. A single life has many different futures, each starting in a different place. Futures starting earlier are larger, and futures starting later are smaller. Larger futures contain smaller ones as parts. Crucially, a future of a life, in my stipulated sense, might include parts of the life which we would now refer to as being in the past. Thus some futures of your life include what you did yesterday. Does that help?
Dale. On your most recent comment. I want to develop a view in population ethics without first settling what the correct theory of welfare is. To do that, I need an account of a life worth living that is independent of any particular theory of welfare. Perhaps I could achieve this by going through every theory of welfare one by one and giving a different account of a life worth living for each. But that would be incredibly laborious. (And I’m not sure it would even be possible. What account would I give for the theory according to which welfare is not analysable?) A better approach, I think, is to give a more general account in terms of something which all theories of welfare have in common, i.e. a ranking of lives from better to worse. I say, give me your ranking, whatever it is, and I’ll tell you which lives are worth living according to your theory.
Campbell,
I do worry that this is a pseudo-problem. You write that:
“We can’t say the two relata are (i) the life with the life and (ii) the life without the life. If you remove the life from the life, you’re left with nothing; one relatum vanishes.
My proposal is a way to overcome this problem. I try to use something like the mathematical concept of converging to a limit. We keep making the future bigger and bigger (and hence the life without the future smaller and smaller) until we get convergence. As Jamie points out, this might make one relatum get very, very small, becoming a “flashlife”. But at least it never vanishes altogether.”
Of course no-one wants to put ‘a life without a life’ to (ii). Rather, what they want to put there is no life at all. Now, you might think that this would be odd because no life at all is nothing at all and thus no relata. Well, this is true if we take the ‘better than’ relation to be an extensional context in which the truth of x is better than y is determined by the reference of x and y and the better than a relation between them. But, it’s not clear why we should do so. Instead we could think of the context as an intensional one. In that case, having no life does not have to be thought of as a referring term as long as the thought of having no life at all makes sense. And, it does seem to make sense if we can think of it as a choice ex ante.
And, I don’t see how the flash-lives are going to solve the problem. We can always take any flash-life and ask whether that is better than no life at all. Thus, if the flashes can be compared with nothing, I don’t see why longer lives cannot.
Jussi, I don’t understand your latest comment. You say:
So what do you want to put there? You say “no life at all”. But what is that? I gather you think it is something, not nothing. But what? Is it the empty set, as in Kris’s suggestion? Is there also such a thing as, say, “no elephant at all”? If so, how does that thing differ from no life at all?
Some philosophers have argued that there are such things as absences (perhaps Roy Sorensen is one of them?), and I don’t want to insist that they’re wrong. But absences are surely controversial. So I’d like to avoid appealing to them if I can.
No – I don’t think nothing is some *thing*; not even a thing called an absense. Not all language works referentially. The relevant question is does the expression ‘no life at all’ express a thought even when it is not a referential expression. I think it does. I can understand it well as it is part of ordinary language I’ve learned to use. If does express a meaningful thought, then the complex expression ‘is better than no life at all’ seems meaningful too.
Think of sentences like ‘John is trying to imagine a round circle’. If you take ’round square’ here to be a referring expression, then either we think that there are round squares or we can say that, oh, but there are no round squares. But, surely, John is trying to imagine something. So, what is it? Trying to look for some*thing* here that is the missing relata seems to me just odd as looking for the thing that is no life.
Thus, this:
“You say “no life at all”. But what is that? I gather you think it is something, not nothing. But what? Is it the empty set, as in Kris’s suggestion? Is there also such a thing as, say, “no elephant at all”? If so, how does that thing differ from no life at all?”
just sounds to me like language gone on holiday.
Campbell’s idea is interesting. But I am with Kris and Jussi in not being bothered by the metaphysics. The reason is that it is very hard to say exactly what a person’s life is. People have lots of different things in mind when they talk about someone’s life. Hurka thinks a life is a sequence of events taking place in someone’s body. But that leaves out lots of stuff we normally think of as being part of someone’s life (e.g. getting married… in fact almost everything interesting about the person). From what Campbell has said, all we know about lives is that they have temporal parts. So I assume a life is some sort of long event. But which one? Until we have a good answer to that, it’s hard to know whether there is actually a problem about having things to compare, and it’s hard to say whether Kris’s idea is particularly problematic.
I am inclined to ditch talk about “lives” altogether, and replace it with talk about the value of a *world* for a person. (Feldman does this I think, but I don’t remember why.) This avoids the problem about having to determine which states of affairs or events count as being part of someone’s life. It also removes the problem about having two things to compare. We’re comparing the subject-relative values of two worlds. The issue then is whether a world can have a value for a person if that person doesn’t exist at that world. I don’t see any reason why not, but it’s something else to argue about.
Jussi,
Regarding this sentence:
I want to say there are two readings of this sentence. On a de re reading, it asserts that John stands in a certain relation, the trying-to-imagine relation, to a certain thing, a round square. But this is surely false, since there is no such thing as a round square.
So we should, if we’re being charitable, give the sentence a different reading, a de dicto one. This reading is harder to explain, but the basic idea, I take it, is that the sentence asserts that John stands in a certain relation to the expression “a round square”. On this reading, the sentence might be true, even though there’s no such thing as a round square.
Now take a different sentence:
I’m not sure that a de dicto reading is available here. What would such a reading say? That John’s life stands in some relation to the expression “no life at all”? I don’t think that could be what we mean when we say John’s life is worth living.
Ben,
Yes, you can avoid the problem of analysing “a life worth living” by not talking about lives at all. But I thought it might be interesting to try to solve the problem, rather than avoiding it. Even if giving a theory in terms of lives is ultimately not the best way to go, it’s still an interesting wee problem to think about. Well, I thought so anyway.
In any case, I prefer your approach to Kris’s. I’d prefer not to talk about lives at all, rather than say that the empty set is a life.
Campbell, yeah, I guess what I said is avoiding the question. But what about this way of answering the question:
S’s life at world w is worth living =df. w is better for S than any world at which S does not exist.
Maybe this would give us what we want.
Some very preliminary results:
Suppose for simplifying purposes that lives come in two discrete chunks, A followed by B. Total lives are L = A+B. Lives have B as a future and, as a limit case, A+B as a future. Suppose also that these lives are given numeric hedonic rankings and that our “better than” calculations are given by comparing the rankings and sums of rankings.
Case 1. A = 10, B = -1. By the proposed definition, B is not a future worth living, since L is not better than A. But neither is A+B a future worth living, since L is not better than L. Since there are no other futures, L is not a life worth living. Intuitively, though, a decent life with a bad blip at the end is worth living (otherwise a lot of us are in trouble). So perhaps we should revise the criterion of a future worth living to say, “L is at least as good as L-F”.
Case 2. A = -100, B = -100. By the revised definition, B is not a future worth living, since L is not better than A. But A+B is a future worth living, since L is always at least as good as L. Consequently L is a life worth living, since every future of which A+B is a part (there are none other than A+B itself) is worth living. Intuitively, though, this life sucks eggs.
More speculatively:
Intuitively, the problem comes in my chunking assumption. In Case 1, we tend to assume that there is some later part of A, call it A*, which is worth (say) 9, such that A*+B is a future worth living.
But if I get to appeal to “organic unities”, there need be no such positively valued A* to redeem B, such that A*+B comes out worth living. The parts of A may get their value only in the context of A itself, so the chunking assumption is in fact not the culprit, I don’t think. (Put more simply: think of my “chunks” as organically united life-stages, such that parts of them would not have nearly the value of the whole.)
Campbell,
I don’t think that that can be right. The Finnish translation:
John yrittää kuvitella pyöreää nelikulmiota.
It asserts the same thing as the English sentence. The truth-conditions are the same and so on. If the English sentence asserted a relation between John and the expression ‘a round square’ and likewise the Finnish sentence between John and the expression ‘pyöreä nelikulmio’, this could not be the case. It’s not that John has any attitudes towards the Finnish expression. The Finnish sentence would come out as false whereas the English true.
Take the sentence:
A bottle full of beer is better than a bottle full of nothing.
That seems very non-problematic. Do we need a relata on the right? What is it? What’s the stuff that this bottle is full of? If there is nothing on that side, the sentence should be problematic. We should need to start looking for a tiny drop or a bit of air. But, it’s not and the sentence can be true even there is literally nothing in the bottle. It just could be that the truth of this sort of sentences does not depend on the success of terms to refer.
Thanks Campbell, that’s much clearer to me. I’ll let F to be a variable taking as values parts of a life L that might begin anywhere in L (except the first instant of L) but that all terminate at t. I assume that lives are closed, so they include their first instants and last instants.
1. A future F of a life L is worth living iff L is better than L-F.
A future meets condition (1) just in case that future makes some positive contribution to the value of your entire life. It is interesting on this account that there will be lots of cases in which a future F of L is worth living without having parts F’ that are worth living. Suppose most of the good things in my life occur very early on. So moments after birth, my future F is worth living. But later on, my life gets more difficult, so my future F’ that is a subset of F is not worth living. So I have a reason to live out my future F and a reason not to live out my future F’. The reverse also can happen. It might be that once I get to F’, everythinng is great and I have a reason now to live out that future. But I might have a good reason not to live out my longer future F. These are cases in which my life starts out pretty bad and gets increasingly better. How do I decide whether to live out the future? Does F’ trump F or the other way around?
2. A life L is worth living iff some future F of L is such that every life of which F is a part is worth living.
Condition (2) is clearer to me. Take the case in which F = L minus one instant (call that L*). Let L* be extremely good and include all of L except its first instant. Given L*, L is clearly worth living. And it seems like it must be the case that at least that future of L is such that every life of which it is a part is worth living. But I’m not sure even that is true. Let L’ be a life longer than L that also includes L*. So L’ includes every part of L except its first instant. Even if L is a very good life, it might be true that L’ is not a good life at all. L’ might be a horrible life. But if L’ is a horrible life, then there is some life of which a future F of L is a part that is not worth living. But now every other future F of L will also be a part of L’. L’ will include L minus two instants, L minus three instants, etc. So there is no future F of L such that every life of which it is a part is worth living. L’ includes every future of L and L’ by hypothesis is not worth living. It follows that L is not worth living. But that cannot be true, since by hypothesis L is a very good life.
Heath,
A few responses:
1. I’m not sure what to make of your cases, because I’m not sure what numeric hedonic rankings are. What does it mean to say, for example, that a certain part of a life has a numeric hedonic ranking of “-1”? In particular, what is the significance, if any, of the fact that this is a negative number?
2. The assumption of two discrete chunks might be a problem for my analysis. As I said earlier, the idea is to find a sequence of futures that converges to limit. But there isn’t much of a sequence with only two chunks.
3. Finally, the analysis is supposed to be consistent with the view that lives are organic unities. Notice that it requires comparisons only between whole lives. L-F is not a part of L; it is a whole life that is a duplicate of a part of L. And although F is merely a part of a life, it’s never compared with anything.
Heath, the whole life doesn’t count as a future. Futures are proper parts of lives.
Ben, I’m worried about the idea of a world being better for a person. This is not the ordinary notion of something being good or bad for a person. (If it were, I should be very excited by the uncountably many worlds that are really, really good for me, and very upset about the uncountably many that are so awful for me.)
Suppose you describe for me a world similar to this one but in which I have x-ray vision, and you ask me whether that world is better for me. To answer, I think about whether the life I have in that world is better than my actual life. But I can’t do that to decide whether a world in which I do not exist is better for me than this world. So, I feel like the original problem has not really been solved.
Jussi and Campbell, I think it’s pretty clear that the intentional object is not a bit of language but the semantic value of that bit of language (its content, say). I don’t exactly understand how Jussi’s idea is supposed to work, though. Are the intensions themselves supposed to be better or worse? I can’t get a grip on what that would be. Also, I would have thought that ‘better than’ was extensional. (If the big yellow fruit is better than the small red one, we can conclude that the grapefruit is better than the cherry as soon as we discover that the big yellow fruit is the grapefruit and the small red one is the cherry.)
Jussi,
Yes, we do need a relatum. It’s an empty bottle.
Sorry, that was a little facetious. But I don’t think we’re making much progress here. Your point about the Finnish translation is a good one. I didn’t get the de dicto reading quite right (as Jamie points out). But I’m still no closer to understanding what you think it means to say “John’s life is better than no life at all”.
Jamie,
I don’t know why you say this: “If it were, I should be very excited by the uncountably many worlds that are really, really good for me, and very upset about the uncountably many that are so awful for me.” How does that follow? Presumably whatever you think about how you should feel about merely possible *worlds*, you should think the same thing about merely possible *lives*. You probably shouldn’t get very excited about either one. Maybe I’m missing something.
You’re right that more could be said about this notion of a world being good for a person. Consider an analogy with time. We say that Jussi is having a good day, or a good year. Another way we might say this is: it’s a good day (year) for Jussi. I take it this is a totally ordinary thing to say. Now think of a world as just a really big time.
Ben,
I don’t think merely possible lives are good or bad for me. I think they are good or bad. They are good to live, so living one might be good for me, but I won’t get excited about that since there is no way I’m going to live one.
The Big Time idea I like better. But when a year is good for Jussi, there are two possibilities. First, in an unproblematic case, Jussi inhabits the year. Second (and I know you’re committed to this anyway and think it’s fine), it might occur after his death, but even then it is spatiotemporally connected to his life, and its events are causally connected. This doesn’t seem optional, to me at least. So a merely possible Big Time doesn’t work in the same way.
Are there any fairly ordinary cases in which we’d say that a merely possible time is good for Jussi? I can’t think of any.
Was any day in, say, the 19th century a good day for Jussi?
Sorry, wrong question. Was any day in the 19th century a worse day for Jussi than today?
I don’t have a fully worked out position on the semantics of ‘nothing’. I guess the point I’m trying to make is that we often compare things with nothing at all and talk of relations to nothing at all.
‘She said nothing at all and that was better than had she revealed the secret.’
‘It’s better to get 10$ than nothing at all.’
‘That was nothing at all.’
‘There was nothing at all on the table.’
‘Those instructions were as good as nothing at all.’
In none of these cases there is a mystery about the missing relata. We understand what is being said here. If other things can be compared and related to nothing, what is so special about lives that they cannot? Why is it that we cannot give the same account we give of the other things we say about nothingness for no life at all? What makes no *life* at all such a special case of ‘no x at all’ that when we put anything else for x we can work with the better than relation but not when a life is in question?
Jamie’s observation about extensionality seems right. I have to think about that more. At least we can talk about better than relations between non-existing things. The truth of a sentence x is better than y does not seem to require that x and y exist – possibilities seem to be enough. Isn’t John not living a life a possibility we can compare to his life?
My account? I did already give it, but here it goes again:
Imagine one had two choices from an ex ante position:
a) To get to live John’s life
b) not to get born at all.
Based solely on the material and social conditions of John’s life, experiantal quality of his life, success his rational aims, amount of autonomy, important relationships, and so on, one has more reason to choose a) than b). I can think of lives for John where this would not be true. I can think of lives of which it would be hard to tell. I’m most certainly mistaken about many lives. But, I cannot see why I could not think of this because b) is nothing at all.
Jussi,
In the sample sentences, ‘nothing’ is a quantifier. Right? So the semantics is not problematic.
Let’s go over them.
‘She said nothing at all and that was better than had she revealed the secret.’
There is a relatum: her saying nothing at all.
‘It’s better to get 10$ than nothing at all.’
Again, the relatum is: getting nothing at all.
Notice that in these cases there is no question of the relatum being the referent of ‘nothing’, which (of course) has no referent (since it is a quantifier).
‘That was nothing at all.’
This one needs more context. I presume it is metaphorical?
‘There was nothing at all on the table.’
Here there is no relatum.
∼∃xON(x, the table)
‘Those instructions were as good as nothing at all.’
This one also needs more context. If taken completely literally, it does seem to suffer the same problem as the one we’re now worrying about.
Jamie,
I think there are all sorts of cases where a merely possible time is good for Jussi. (Maybe it’s more natural to say that those times *would be* good for Jussi, if they were actual. Does that help?) Every good moment of every possible life of his is such a case. We imagine Jussi having superpowers for a while, flying around being happy. That time is (or would be) good for Jussi.
Anyway, it’s true there will still be something to argue about here. I say my solution solves the problem Campbell was worried about, of not having two things to compare. But you may still want to say that even though we have two worlds to compare, one of them lacks the relevant property (the subject-relative value). Then we have to argue about what it takes for a world or a time to have a value for a person. Does the person have to exist at that world or time in order for the world or time to have that property? I say no. I haven’t given any argument for that claim, though. I also don’t know of any good argument that the answer is yes.
Campbell,
In answer to your last question, I say yes, as long as Jussi is enjoying himself today, a day in the 19th century is worse for Jussi than today is. (Unless Jussi is way older than I think.)
Ben,
Well, I like the ‘would be’, but I hate the ‘if they were actual’. For each merely possible time, it is impossible for it to be actual! For something to be actual is for it to exist in this world. You know, @. That one. This one. And worlds have their contents (elements, constituents, etc.) necessarily.
So we need a different antecedent. That high-paying job would be good for Jussi, if only it wouldn’t go to his head. That sweater would be good for him, if only it fit.
Okay, those aren’t good models. This is better:
And the year of living dangerously would be good for him if he were to live it.
But now, the world w, the one in which he never exists, would be good for him, if… if only… what?
Ben, your answer doesn’t seem like a “totally ordinary thing to say”.
Jamie,
Well, on my view the reason your last question can’t be answered is that there is no world such that Jussi doesn’t exist there, and such that it is good for Jussi. Every world where he doesn’t exist has zero value for him, because nothing good or bad happens to him there. We might, however, say that a world where he never exists would be better for him than the actual world, if things are actually going badly for him. At least, I’m OK with that. I’m waiting for an argument that it’s wrong.
Seriously though, nothing turns on the fact that a merely possible world can’t be actual, does it? The business of axiology is to put values on these things, whether they are actual or not.
Campbell, no, it doesn’t seem ordinary. I only said it seems OK to say that some times are good for some people. With some pairs of people and times, we don’t normally say that the time has a value for the person. There are lots of possible reasons for that.
Ben, sorry, I don’t get your answer.
I was saying that I was happy with the thought that a time (a year, say) that is merely possible would be good for Jussi, if… something. And I am asking what the something is, in the important case. I gave examples of what the something (the antecedent) is in ordinary cases.
I agree that axiology puts values on things, actual or otherwise, of course. I agreed with you that possible lives have value, for instance. But at the moment we are talking about some merely possible things (years and worlds) being good for a person, and trying to make sense out of that. Here’s one way. A merely possible life is good for you, we might stipulate, iff it would be good for you to live it. This is an acceptable stipulation. (I would not put it that way. I would say that the merely possible life is simply good, not good for you or me.) However, this, I pointed out, doesn’t seem to help with the case of a merely possible world being good or bad for me, or better or worse than another for Jussi. We need some other would, and I was asking what it could be.
So that’s where we stand, I believe.
“Seriously though, nothing turns on the fact that a merely possible world can’t be actual, does it?”
If a merely possible world couldn’t be actual, in what sense is it possible? Something has gone wrong with the discussion about possible worlds.
More importantly, people do say things like “it would be good for me if x happened” or “things would go very badly for me in that situation” or “Things would be neutral from my perspective if that were the case.”
We understand the idea that a situation can be good, bad, or neutral for a person. One and the same situation might be good for one person but bad for another.
It’s not obvious to me that we have to understand or analyze this notion in terms of lives persons could live. We could just take the notion as undefined. What’s wrong with doing that?
If we take the notion of a situation having value for a person as primitive, then it seems pretty clear how to define up the notion of a world having value for a person, since one could take a possible world (at least for axiological purposes) as being a maximally consistent situation.
It’s then a substantive question (as Ben suggested) whether a world that one does not exist at can be good, bad, or neutral for a person. I’m inclined to think that it can be neutral, and accordingly, if the actual world is bad world for you, worlds in which you were never born are better for you.
Well, here is how I thnk possible worlds talk works.
For something to be actual is for it to be in (part of?) the actual world. Call the actual world ‘@.’
Something could be the case: it is the case in some world.
Take some merely possible world, w. Consider:
w could be actual. (Or, could have been, if you prefer.)
So that’s true iff: in some world, w is @. But there is no world in which w is @.
Right. And I agree that it’s fine to think of a ‘world’ as a maximally specific situation. But in your (ordinary talk) examples, you, the person for whom the situation would be good or bad, exist in the situation that would be good or bad. I agree that this seems perfectly comprehensible. (I would like an analysis, but we don’t need one to be justified in supposing that the things make good sense.)
But, to me, the idea of a situation being good or bad for me, when in that situation I never have existed nor ever will exist, is not clear at all. We are speaking of worlds being good for people in other worlds. I can’t understand that, so I need an analysis.
I’ll very probably now allow Campbell to take up the position, having said about all I can.
If a merely possible world couldn’t be actual, in what sense is it possible?
As far as I can see, there’s exactly no sense in which such a world is possible, as Kris implies. It is true at each world w that w is actual (whatever modal metaphysics you happen to find congenial). So if w is possible, then w is possibly actual. There are no worlds that are possible but not possibly actual.
There might be worlds that are possible, but not feasible, but that depends on some Plantingan assumptions not everyone accepts. In any case, even if you do accept those assumptions, even worlds that are not feasible are possibly actual.
Some “two-dimensionalism” might be helpful here. (No, I’m not going back on my earlier four-dimensionalism. Different dimensions.) Very briefly, in 2D modal semantics there are two ways to think of possible worlds: we can think of them either “as counterfactual” or “as actual”. Corresponding to these are two notions of necessity: truth in all worlds considered as counterfactual, and truth in all worlds considered as actual. In the former sense of necessity, “not actually p” implies “necessarily not actually p”. This is the sense Jamie has in mind. But in the latter sense, that implication is invalid. This may be the sense others, Kris and Mike, have in mind.
Campbell, I never understood what “considered as actual” means. It’s as if you told me to consider Frank Jackson as me, or to consider 6 February 1840 as today, or Grosvenor Square as here. I know how the logic works, I just don’t know how to interpret it.
Corresponding to these are two notions of necessity: truth in all worlds considered as counterfactual, and truth in all worlds considered as actual. In the former sense of necessity, “not actually p” implies “necessarily not actually p”.
I don’t think I denied that inference, which I’m sure is valid. It has to be right that “it is not actually raining” (asserted here at our world) entails that, at any other world, it is true at our world that it’s not raining. I’m not sure how that inference comes out invalid if I read necessity in the second way. Name our world @. I consider world w as actual and ask whether the proposition expressed by ‘it is not actually raining’ as uttered here (in @) is true there (in w). But the proposition expressed by that sentence is “it is not raining in @”. So it looks like that ought to be true at w, whether or not it is raining in w.
In any case, I was denying that there are possible worlds that are not possibly actual. I’m not sure how this is addressed in the move to 2D.
Mike,
Well, here is the inference that you agree is valid:
¬Ap ⇒ ¬◊Ap
So take a merely possible world w and let p be the complete conjunction of truths at that world (so p’s truth is w’s existence). Since w is not actual,
¬Ap
So, by the valid rule,
¬◊Ap
So it is not possible that w be actual.
Hi Jamie,
This is interesting, but I’m sure misleading. Let’s agree that worlds just are maximally consistent sets of propositions. So, let p = w. Let the actual world (our world) be @.
1. ¬Ap ⇒ ¬◊Ap
(1) says that if w is not @, then it is true in every world that w is not @.
2. ¬Ap
(2) says that w is not @
3. /:. ¬◊Ap
(3) says that it is true in every world that w is not @
The inference is valid, no question. But what it shows is not that w could not be actual. If it did show that, then we could show that every non-actual world is necessarily non-actual. Therefore there is one possible world! But that is certainly not true. In fact, any possible world could be actual. What the proof shows instead is that w is not identical to the world that happens to be actual. Indeed it is necessarily not identical to the world that happens to be actual (viz., our world here).
I agree with that entirely. You could put it in terms of what propositions hold at our world. Put in those terms, the conclusion says that necessarily the maximally consistent, conjunctive proposition p does not hold at our world. It does not say that the maximally consistent conjunctive proposition p couldn’t be actual.
It’s nice to see that my life and its counterparts get so much attention…
Jamie,
A very late thought about this:
“I don’t exactly understand how Jussi’s idea is supposed to work, though. Are the intensions themselves supposed to be better or worse? I can’t get a grip on what that would be. Also, I would have thought that ‘better than’ was extensional.”
I’ve thought about this. You are right that ‘better than’ cannot be a relation between intensions. And, I think it’s true that normally the expression refers to a relation between objects and thus is extensional.
This is going to be maybe a bit of ad hocery. But, perhaps one could think that ‘the better than’ bit works differently when we talk about better than no life at all than when we talk about being better than other things. You could think that ‘no life at all’ has an intension which obviously lacks on extension the complex expression ‘is better than no life at all’ has a compositional intension that determines an extension to which many lives belong. That compositional intension would be in part a consequence of the thought expressed by no life at all.
This would be to take the expression to be, contrary to appearances, a one-place predicate with its own ‘principle of classification’. It would also be an extensional context – so substituting corefferring terms to a life would not change the truth-value when this complex predicate is used. We could then have a debate about what is the correct way to characterise the epistemic means with which we pick out lives with this predicate. My suggestion is one but I’m sure there are better ones. This would save us from looking for the missing relata.
The worry with this is that we would have to think that ‘being better than’ is used in somewhat different metaphorical meaning in the expression. But, if this saves the intuitive expression from the metaphysical oddities, I might be willing to pay the price.
If we insist that is better than is an extensional relation, and that it must have a relata in the case we are talking about, the solutions I suggested, one of which involves taking something like the empty set as a life, looks less ad hoc or arbitrary or metaphysically suspicious — especially if you are already inclined to take the lives discussed in axiology to be complexes of propositions or states of affairs anyways (as I am.)
True. But, you would have to say something about the existence of sets. Do they really belong to the ontology of the world? Even empty ones? I’m also less than sure how an empty set could be a life. The former seems like an abstract object and the latter something more concrete. And, comparing lives to abstract objects seems like comparing whether apples are better than modus ponenses.
Well, Mike, “Ap” says that w is actual, we’ve agreed about that. And you agree that we can prove ¬◊Ap. That means we can prove that it is not possible that w be actual. But you insist that it is possible that w be actual.
I think I know how you got into this tough spot. You think that (i) every non-actual world is necessarily non-actual entails (ii) there is one possible world. If so, that would be a good reason to reject (i). But it’s a fallacious inference. It’s like inferring there is only one day from no day besides today has ever been or ever will be today.
I don’t think (i) entails (ii) at all, and never suggested it. Obviously, if I thought that, I would never have affirmed it.
(i) every non-actual world is necessarily non-actual entails (ii) there is one possible world.
On the contrary, I’ve been reading Campbell’s principle in the only way it can come out true. The only way that comes out true is as stating that every world that is not identical to the actual world, viz. @, is necessarily not identical to it.
And you’ll notice that this is exactly how I read it above at July 28, 2007 at 09:23 AM.
1. ¬Ap ⇒ ¬◊Ap
(1) says that if w is not @, then it is true in every world that w is not @.
Read in the way that I render it, (1) is part of an argument that is valid, as I said above. But the argument does not show that w could not be actual and I say so explicitly.
The inference is valid, no question. But what it shows is not that w could not be actual. If it did show that, then we could show that every non-actual world is necessarily non-actual.
So yes I affirm Campbell’s principle, but that does not commit me to the view that there are possible worlds that could not be actual. As far as I know, you’re the only one defending the view that there are possible worlds that could not be actual. I think that claim is necessarily false.
Jaime,
I think maybe the following will help dispel the disagreement, if there still is one. I am using ‘actually’ in the same way that most everyone uses ‘here’. I think that is the common use of ‘actually’. So, in (1) ‘here’ refers to the location of utterance.
1. It is raining here.
But ‘here’ continues to refer to the location of utterance at other locations. So you look outside and see it pouring rain and ask ‘I wonder if it is true in Barcelona that it’s raining here’? Of course, it is true in Barcelona that it is raining here, since ‘here’ does not vary with the location of evaluation. It continues to refer to the location of utterance. I am using ‘actually’ he same way. So taken rigidly, these are both true,
3. If it is raining here, then it is true everywhere that it is raining here.
4. If it is actually raining outside, then it is necessarily actually raining outside.
But if you do not take these terms rigidly, then both (3) and (4) are false. But as I said ‘actually’ and its cognates are not always rigid. It is not rigid in (5) or (6), since (5) is contingent, and (6) is false if ‘actual’ is read rigidly.
5. @ is actual.
6. w is possibly actual.
So this probably explains the disagreement. Maybe you don’t think the natural reading of (4) is the rigid reading. Maybe you don’t think that ‘actually’ is even typically used this way. Maybe you think (6) is false because in (6) you are reading ‘actual’ rigidly. I don’t know. But the explanation, I’ll bet, is in the vicinity.
Mike,
Hm. You wrote this:
That every non-actual world is necessarily non-actual, that’s (i). That there is one possible world, that’s (ii). So there, at 9:23, you wrote (i), followed by ‘Therefore’, followed by (ii). That’s why I figured you believed that (i) entails (ii).
I’ll get to the rest later, I guess.
Right, I should not have phrased the claim that way. What I was asserting was, as I frankly think is evident, if your argument showed what you believe it shows, then it would be true of every possible world that it could not be actual. That’s the absurd consequence of this argument. Hope that clarifies matters there. Still, I should not assume we’re using blogging precision. I’ll be more careful.
Now, it would be great to learn, in addition to this, which possible worlds could not be actual. Which possible worlds are those, exactly? I’d understand entirely, I guess, if you didn’t get to it.
I thought the ‘absurd consequence’ was that there is just one possible world. That is absurd, but, we both now agree, it is not a consequence. You now say the ‘absurd consequence’ is that none of the merely possible worlds could be actual. But that is a theorem of modal logic with an Actually operator. So I don’t get the objection.
‘Actually’ is rigid. Your (5) is not contingent but necessary. I agree entirely that ‘actually’ works like ‘here’. It is a true indexical, and like other true indexicals, it is rigid. I wonder whether you think that ‘here’ is also sometimes non-rigid — I also wonder what ‘Actually’ means when it isn’t rigid. (I found in Scott Soames’ book an alternative operator, Actually*, but he says that one is also rigid.)
“Ap”, we agree, says that w is actual. “¬◊Ap”, then, says that it is not possible that w is actual. So, the conclusion of the argument says that w could not be actual.
You mean, the only one here at PEA Soup? Maybe. Kaplan, Soames, and Kripke all accept it, though, so I’m not too worried.
Jamie,
Again, the uses of ‘actuality’ are getting confused. Obviously, I agree that if ‘actuality’ is read rigidly, then, it is true that,
1. No merely possible worlds could be actual.
But as I have been at pains to say, this is the trivial claim that no merely possible world is identical to the actual world. I agree that everyone believes that. But certainly that is not what you were arguing for. You weren’t arguing that @ is the actual world and w is a merely possible world and necessarily, w ain’t @, were you? Instead, I thought you were arguing that (1) is true with ‘actually’ read non-rigidly. That is the claim that I was worried about (in fact, Kris had that worry before I did. I just seconded it.). I’m pretty sure that Kaplan and company don’t think that’s true.
Elsewhere you say,
‘Actually’ is rigid. Your (5) is not contingent but necessary.
My (5) says this,
5. @ is actual.
In (5) the term ‘actual’ is most naturally read as non-rigid. See David Lewis, Philosophical Papers I p. 22, propositions (1)-(5) on the shiftiness of ‘actuality’ and A. Plantinga The Nature of Necessity p. 50, proposition (6). As Lewis points out, ‘actuality’ has both a shifty (non-rigid) and a shifty sense. We need both, since in some sentences ‘actuality’ is used rigidly and in some sentences it is not. But both can express true propositions.
Now if you read (5) in a non-shifty way, it is true that the actual world is necessarily actual. But that is not the natural reading of (5). It is more naturally read as a contingent proposition in which ‘actual’ is shifty, or non-rigid. Certainly we want a sense of ‘actual’ on which it comes out true that, if @ is the name of our world, @ might not have been actual. That won’t come out true of ‘actual’ is always and everywhere rigid.
Ahhhhh.
Okay, so what does ‘actual’ mean in its non-rigid use? I just took a quick look at the Lewis paper, and there the ‘secondary’ sense seems to be transparent, pleonastic, redundant, or something like that. (I mean, the truth conditions don’t change when you just remove it.)
Here are two quick points. First, I think you might want Campbell’s second kind of ‘necessity’. Campbell noted that in the first kind, but not the second, a certain inference is valid. (I find this useful, because as I said I understand the logic, but not the semantics, of the second kind.) You said you accepted the inference as valid. But if it is, one simply can’t avoid the conclusion, for each merely possible world, that it couldn’t have been actual.
Second, I think things may stay clearer if instead of picking out worlds by names or other denoting expressions, we use maximally specific propositions (so, once again, the truth of the proposition will amount to the existence of its world).
Put those two together. Here’s the inference rule in question, again:
¬Ap ⇒ ¬◊Ap
It’s completely general, and particularly relevant when the proposition p is one of the maximally specific ones.
Jaime,
That’s fine. I agree that the inference rule you describe is true for ‘actuality’ understood rigidly. In fact, I agree that that is how the actuality operator works (I think ‘A’ is an operator in the rule). No problem. I also agree that we can arrive at the conclusion that ‘no merely possible world could be actual’ so long as we are using ‘actual’ rigidly. I deny, and I’m guessing you do as well, that the claim is true on a non-rigid reading.
When Lewis says that ‘actual’ is used in a shifty way, I take it that he means that the reference of the term ‘actual’ shifts to each world of evaluation. It doesn’t continue to refer to the world of utterance. So, if I say, (1), with the shifty ‘actual’,
(1) Necessarily, actually it is raining.
then (1) is true only if at each world w, it is raining at w. I evaluate the sentence ‘it is actually raining’ at each possible world w with ‘actually’ shifting reference to each world w of evaluation. If there is some world at which it is not raining, then (1) is false.
‘Actually’ is redundant, I take it, in the sense that there is no difference in meaning between (1) above and (2) with ‘actually’ removed. They are true or false at the same worlds.
2. Necessarily it is raining.
Good. So that (secondary) use is redundant.
Now, back to Ben’s idea.
We say, ordinarily and apparently unproblematically, that this or that situation (or perhaps time, meaning some period of time like a year) is good for a person (say, Jussi). Now I want to know, what does it mean to say of a merely possible situation that it is good or bad or better or worse for Jussi?
One idea was that to say that it is better is to say that it would be better. Would be, if what?
First, an ordinary example. Living in Houston would be better for Jussi if he loved hot, steamy weather. Now a difficult example. The situation in which Jussi never has existed and never will exist would be worse for Jussi if… what?
I thought the next suggestion was: if that situation were actual. But now we see that this can’t be right. For it either means something that is impossible, or else the ‘actual’ is redundant. So it comes out: the situation in which Jussi never has existed and never will exist would be worse for Jussi if Jussi never had existed and never would exist. I find this incomprehensible; the antecedent doesn’t help at all, we have made no progress.
I have a lot of sympathy with the general strategy of refusing to let a philosophical theory tell us what does or does not make sense, so I think it’s worth pursuing the idea that comparison of non-existence with some real life turns out to be unproblematic if only we’ll drop some hard-to-notice theoretical assumption. I just don’t yet see how it will work.
Jaime,
I’m not even sure the easier situation is so easy.
Living in Houston would be better for Jussi if he loved hot, steamy weather
It is not clear to me whether the situation would be better for Jussi if he loved hot weather, since were he to live in Houston his preferences might be different. Do you evaluate how good the counterfactual situation is by appeal to the counterfactual preferences or by the actual preferences of Jussi? I remember Wlodek Rabinowicz somewhere arguing that is should go by actual preferences. But I don’t know.
About the situation in which Jussi never existed, I think we could say (dropping the ‘actual’) that Jussi will have an on balance good life and were it the case that Jussi never existed, it would not be the case that he had an on balance good life.
W. If S is on balance good and S’ is not on balance good, then S’ is worse than S.
I think it’s true in the counterfactual situation that it is not the case that the value of Jussi’s life is on balance good, since it is trivally false that Jussi’s life is on balance good there. It is also trivally false that Jussi’s life is on balance bad, but that might not matter if (W) is true.
I have this feeling that the discussion is getting hung up on accidents pertaining to english, and that it might be my fault. Let me add a couple of remarks.
Forget about well-being and just think about (agent-neutral) value for a second. We say things like this: “it would be better if X happened,” where X doesn’t actually happen. That’s the normal way to speak english. Surely all we mean by this is: a world where X happens – call it Wx – is better than this world (@). We *don’t* mean this: Wx *would be* better than @. At least, if we say such a thing, we don’t implicitly think that there’s an ‘if’ clause missing from the end of the statement. We’re not saying that under some circumstance we’re imagining, Wx is better than @, but it’s not actually better.
The exact same thing is true of well-being. We say things like ‘Wx would be better for Jussi than @,’ but we just mean that Wx *is* better for Jussi – not that it would be better, under some circumstance that needs to be plugged in for the statement to make sense.
The exact same thing is true of well-being. We say things like ‘Wx would be better for Jussi than @,’ but we just mean that Wx *is* better for Jussi
Ben, you say ‘Wx’ is a world where x happens. But there are lots of worlds where x happens. Certainly some worlds where x happens are better for Jussi and some are not. It would be pretty rare that every world where x happens is better for Jussi than any world in which x doesn’t happen. So we might want to distinguish the closest worlds where x happens from the rest of the worlds where x happens. It might be that the closest worlds to ours in which x happens are terrible for Jussi, but some distant worlds where x happens are great for Jussi.
When we talk about it being better for Jussi, were x to happen, I think we mean only that the closest worlds where x happens are good for him (not every x-world). Am I following you?
Mike, the thing is, it would be nice to be able to analyze and explain “on balance good” in terms of comparison to no-life, or comparison to how good a world is for a person who isn’t in that world (the latter is not put well, but I mean Ben’s approach).
Ben, somehow this isn’t helping. The better-for-me relation looks very weird when I do not exist in one of the relata-worlds. Compare: the situation in which I am surrounded by great white sharks is very scary for me. The situation in which I am surrounded by fluffy white kittens is not scary for me at all. The situation in which I do not exist… scary-for-me doesn’t seem to apply to that situation. When a situation is scary-for-me, that is to say it’s scary to be in. When a situation is good-for-me, that is to say it’s good to be in.
But I should think about this more, for sure.
Mike, right, that’s what I meant.
Jamie, sure, it’s still an option to say that worlds where you don’t exist have no value at all for you (not even zero). For all I know that might be the majority view about the subject. I just felt like at some point, some other issues came in – about modal metaphysics and such – that were not relevant to that question. I think there are pretty good arguments for assigning those worlds a value of zero for you. But I’ll hold off for now.