You are offered a choice between two experience machines.
Machine A is just like the ones you already know about. You’ll be on the experience machine until you are 120 years old, getting lots of various sorts of pleasure.
Machine B offers exactly the same experiences as Machine A. However, the experiences are crammed into a much shorter period of time: just one day. The first 12 hours of this day seem to last 12 hours. The next 6 hours seem to last 12 hours. The next 3 hours seem to last 12 hours. And so on (as in Sorensen’s "The Cheated God"). A fraction of a second before the 24 hours are up, the machine kills you painlessly. At that time it seems to you as if you have lived 120 highly pleasant years.
Which machine should you choose?
It seems to me that a hedonist should be indifferent between these two machines. Life on one is experientially indistinguishable from life on the other. But suppose we attempted to get an objective measure of the amount of pleasure you get in each machine. The pleasures in Machine A last a lot longer than the pleasures in Machine B. So objectively speaking, Machine A offers more pleasure.
I guess some hedonists might prefer Machine A. But the example at least shows that there should be two ways for a hedonist to account for duration. An "objective hedonist" would look at the actual durations of pleasures, in absolute time. A "subjective hedonist" would look at how long the pleasures last in the personal time of the subject. I think more needs to be said about how to formulate hedonism in terms of personal time, but I thought I’d throw the idea out there and see what people think.
The experience of pleasure lasts 120 years in both machines. The difference between Machine A and Machine B is that only in the former do the physical processes on which those experiences supervene last 120 years as well. Since hedonism assigns intrinsic value solely to experiences, hedonists should be indifferent between the two machines.
Note that if you wanted to preserve space for an objective hedonist to be indifferent between the machines, you might deny that Machine A’s increased duration results means that it provides a greater *amount* of pleasure. Even in objective terms, the amount of pleasure in an experience isn’t *just* a function of its duration, right? Maybe the increase in personal time is another way for the pleasures to be proportionately more “intense,” for example.
On the other hand, along with Pablo, I don’t see any reason why hedonists ought to be interested in absolute, rather than personal, time at all.
Thanks for the comments, Pablo and Benjamin.
It had not occurred to me to say that the experiences themselves last 120 years in both cases. In Machine B, the experiences *seem* to last 120 years; but the way I was imagining the case, they really don’t last that long.
There may be different ways to get a machine like B. One would involve time travel: after 18 hours, you travel back in time 6 hours; etc. etc. On such a machine, it’s easy to see how it would make sense to say that your experience really lasts 120 years: the parts of the experience temporally overlap. The sort of machine I was imagining was a machine where your experience really doesn’t last that long, it only seems to.
I guess I think it would be strange, also, for a 120-year event to supervene on a 1-day event (assuming there is such a thing as supervenience between individual events). Pablo, can you help me understand this?
If we are thinking of the intensity of a pleasure as something like the number of hedons per minute, then Machine B does provide more intense pleasure. That’s a cool idea. I’m not sure this is an intuitive way to think about how intense the pleasures are in Machine B.
Doesn’t this all depend on how we identify “pleasure”? For instance, let’s say you identify pleasure with the qualitative experience of pleasure; in other words the “seeming”. So in that case, there is no sense in which, “objectively speaking”, Machine A gives you more pleasure. The qualia in both cases are identical, despite one’s being compressed in time, aren’t they?
I guess you might be able to differentiate the machines by identifying pleasure by some “objective” state, like a brain activation state, or something. In which case, that brain activation state would genuinely last 120 years in Machine A. But I’m puzzled as to how to interpret Machine B on such a view. On that view, you would just deny, I think, that there was anything relevant about the subject’s so-called conscious experience; the subject had 1 day of that brain activation state. Machine A is clearly better. Or did you have something else in mind? Insofar as I identify pleasure as some qualitative state, I’d say “indifference.”
Hi Dale,
What we think pleasure is might well have something to do with what we should say here. But I’m not sure identifying pleasure with qualia will help. Is the idea supposed to be that there’s no appearance/reality distinction with qualia? That doesn’t seem right. We can be wrong about how long our qualia last. That’s what I’m thinking happens in Machine B. The qualia last only a day but seem to last 120 years.
Ben,
Suppose at noon Alma is hooked up to Machine A and Bob is hooked up to Machine B.
In the period from noon until midnight, Bob and Alma experience the same amount of pleasure, plainly. What about in the period from midnight until 6 am? In that period, does Bob experience as much pleasure as Alma does?
I don’t have strong intuitions about the case, myself (I’m not a hedonist so maybe that’s why). But it does seem to me that if Bob experiences as much pleasure as Alma in the midnight to 6 am period, then a hedonist ought to think that Machine B is exactly as good (to be hooked up to) as Machine A. But if Bob experiences less pleasure than Alma does, then it’s not as good.
(This point is closely connected to Benjamin Bagley’s, obviously.)
Right. I wasn’t trying to insist that the experience lasts the same amount of actual time. Rather, I was only noting what you say, that the qualitative experiences are indistinguishable. And if you identify pleasure with a qualitative experience, (I think) you’re going to say that the fact that one takes place in a day, the other in 120 years is simply irrelevant; the qualitative experience is the same, hence the amount of pleasure is the same, hence, if one is a hedonist, the value is the same. (In other words, I don’t see any motivation for suggesting that A is better if one defines pleasure as some qualitative experience.)
Jamie,
I think I see what you’re driving at. But I think you meant to say that if Bob gets *twice as much* pleasure as Alma from midnight until 6pm (by getting pleasures that are twice as intense, as Bagley says), then Machine B is just as good. Right?
Dale,
I’m not sure if we disagree or not. I’m thinking about the possibility that, although the experiences are indistinguishable to the people experiencing them, they might in fact be distinguishable. Machine B is producing some extra deception in the subject, concerning how fast time is elapsing. So the subject is having experiences that are in fact pretty short, but seem long. If that’s what is going on, then there’s conceptual space to say that the life on Machine A would really be better, even though the subjects couldn’t tell the difference. I think that hedonists shouldn’t care about such unnoticeable differences, but maybe some would.
Argh, yes, that’s right. Or, as much as Alma gets from midnight to the next noon, I guess.
I was thinking that questions about the intensity of pleasure are a bit more difficult than questions about whether one person gets more than another over certain durations. I have a better grip on those, myself, in any case. I would have said that Ben must get the same amount of pleasure from midnight to 6 as Alma gets from midnight to noon. If this is right, then something like Benjamin’s suggestion must also be right (I think).
Ben,
You write: “Machine B offers exactly the same experiences as Machine A. However, the experiences are crammed into a much shorter period of time: just one day.”
But I wonder whether these two claims are compatible. Of course, I know what you mean by the first claim: Machine B offers you experiences that seem to you to be exactly the same as the ones Machine A offers. But from the fact that they seem to you to be the same, I don’t think it follows that they are the same.
Consider Machine C. Machine C gives you exactly ten seconds of experiences and then kills you. But part of what Machine C gives you is the apparent memories of x years of pleasure (where x is 120 minus your current age, I guess) of just the sort that Machine A would provide (and just the sort that Machine B would seem to provide). So Machine C, like Machine B, offers you experiences that seem to you to be exactly the same as the ones Machine A offers. Being hooked up to Machine C, that is, is indistinguishable from being hooked up to Machine A (or B). But I would say that it is false that Machine C offers experiences that are exactly the same as the ones Machine A offers. The person on Machine A experiences a lot more pleasure than the person on Machine C experiences. It’s just that the person on Machine C seems to remember having experienced just as much.
I would definitely choose Machine A over Machine C. And I think I would choose Machine A over Machine B as well. I’m not certain, I will admit, that Machine B is inferior to Machine A in the way that Machine C is (i.e. that it offers less pleasure). But I’m not certain that it’s not inferior in this way, either—and I’d rather not take the chance.
Is it supposed be significant that one’s experiences in Machine B are, so to speak, speeding up? Would it make any difference if the rate were constant, i.e. every hour seems to last five years? Just curious.
Hi Ben. You write:
I believe that the duration of an experience is an intrinsic property of that experience. I also believe that phenomenal experiences have all the intrinsic properties they seem to have. Because I have these two beliefs, I also believe that, if some experience seems to last a certain amount of time, then it lasts that amount of time.
(The claim that both machines induce experiences that last 120 years introduces unnecessary complications. I should have said that both machines induce experiences that last the same amount of time. There are also difficulties having to do with our fallibility, demonstrated by Daniel Kahneman, in recollecting and aggregating past experience.)
Incidentally, the issues introduced by this interesting thought experiment are not entirely abstract. Consider that, as people grow older, time seems to speed up.* On my view, this would imply that, on average and other things equal, biological life-years contain experiences of diminishing marginal duration, and hence progressively few of the things that, for the hedonist, make people’s lives go best.
—
* I only know anecdotal evidence supporting this claim, though I have asked a number of friends and acquaintances and the older they are the more frequently they agree with it. See also the informal survey in the second episode of Michio Kaku’s four-part documentary on time (the relevant section begins at 3:23).
Hiya Ben,
If the Subjectivist wins then here’s some reasonable advice: Make your life as boring as absolutely possible…make boring love, eat boring food, reread, I don’t know, a complete history of the lives of silkworms for the rest of your life and all so that your life seems to drag on and on and on. But keep it, your life, at least worth living. And don’t, don’t ever have a fleeting, though moderately pleasurable experience.
Pablo,
You write:
I believe that the duration of an experience is an intrinsic property of that experience. I also believe that phenomenal experiences have all the intrinsic properties they seem to have. Because I have these two beliefs, I also believe that, if some experience seems to last a certain amount of time, then it lasts that amount of time.
But isn’t my Machine C, described in my earlier comment above, a counter-example to this set of claims? Is it really plausible to assert that the person who dies ten seconds after the experience begins has had an experience that lasted 120 years, just because, due to the quasi-memories that filled his head, it seemed to him that he had had a very long experience?
I think it would be more reasonable to take machine B. Even though the experiences seem to be the same there is one important difference. Doing Machine A I will be 120 years old when I get off and on Machine B I will be only one day older then I am now (61). I will probably die immediately after getting off machine A (or wish I had), but will still have some more time to be hooked up again to Machine B. Assuming that I live to I’m 120 then I would have many more times the experinces on B then I could have on A. Where do I sign up?
Do we need the experience machine? I was wondering whether this could *actually happen*. I’m not sure how this works because my theory of relativity isn’t probably the standard one. But, as far as I remember, time goes slower the closer you get to travelling to speed of light. So, imagine then someone on a huge space-ship that travels close to speed of light. Someone on that ship could live a life that was pretty much identical to our own in all respects. But, where our life takes about 70 years or so, theirs could take thousands of years. But, I take it that the life for them wouldn’t be any different. Did I get this completely wrong? Would their longer-lasting pleasures be better?
Come to think of it – if we take the perspective of the space-traveller, the normal Earthling-lives could literally take place within a day. Would the space-traveller be inclined to think that he would therefore rather not have an identical life on Earth? Objective time seems to be a rather relative notion.
Jussi,
I’m no physicist, but wouldn’t you have to say that while their life takes thousands of years in our frame of reference, it only takes 70 or so in theirs?
I’m still fuzzy on the difference between absolute and personal time. I know that dreams are said to happen much more quickly than they seem to, for instance, but I’m frankly a little unclear about what that means. I suspect that it only means that when we reflect back on the dream after waking, our recollection of how long the dream lasted is systematically wrong. If this is right, and if in general “personal time” is primarily a matter of inaccuracies in our memories of how long events lasted, then I’m thinking that Machine A is superior from a hedonistic point of view. It might be that a person on Machine B, nearing the “end of the ride,” would be entirely convinced that she had lived a full human life. But that doesn’t mean that she would in fact have had the same experiences as someone on Machine A. It would just mean that, looking back, she wouldn’t be able to remember what the experiences that she had had were actually like.
I’m sure this isn’t important, but I think your description of Machine B is inconsistent. Your description seems to imply the following: for any positive integer n, if it seems that 12*n hours have passed, then actually 12*(2 – 2^(1-n)) hours have passed. In the case where n=10 we have: if it seems that 120 hours have passed, then actually 23.98 hours (approx.) have passed. But that’s not ‘a fraction of a second before the 24 hours are up’, as you say. It’s closer to 1.4 minutes.
Oh, how embarrassing! I see my mistake already: it should be 120 years, not hours. Please ignore my last comment.
Campbell is providing some evidence for an arithmetic corollary of Hartman’s Law of Prescriptivist Retaliation, to wit:
Any article or statement about correct grammar, punctuation, or spelling is bound to contain at least one eror.
Well, to be fair, my arithmetic was okay. But I took the wrong case. The relevant case is not n=10, but rather n=876600 (allowing for leap years) — admittedly, quite a large discrepancy!
OK, I’m falling behind.
Troy,
You said, “But from the fact that they seem to you to be the same, I don’t think it follows that they are the same.” I think that’s right. I was thinking that the hedonist should say that if two experiences seem just the same to the person experiencing them, they are equally valuable for that person – even if one is really longer than the other.
I want to say there’s an important difference between Machines B and C. Machine C doesn’t give you experiences that seem just like the ones you get in B and A. It only gives you memories of those experiences that seem just like the memories you’d have in B and A. So Machine C is way worse, even though when you were on it you’d think you’d had a life that was really great. The hedonist has to be able to say that we can be mistaken about how pleasant our life has been. On Machine B, you’re being deceived, but it’s not clear that you’re making a mistake about how much pleasure you’re getting.
Campbell,
No, it doesn’t matter that it speeds up. The speeding up only matters if you want to get an apparently infinite life into a finite time. Still, speeding up is cooler.
Christian,
LOL. Right, boring and painful experiences seem to last longer than pleasant ones. More evidence that there is no God. (Machine B is supposed to make things work the opposite way.) I’m pretty sure you’d get a net loss of hedons if you tried your strategy. Do some experiments and let me know how it works.
John A.,
Get in line.
Jussi,
Right, you could give cases that don’t involve experience machines. But in the sorts of cases you’re talking about, it’s just not clear to me how fast time is really passing in the vicinity of the person getting the pleasure. You could also imagine a “Matrix” type case where the stuff around you is moving slowly, so that you could sit down and eat a sandwich and take a nap during the time it takes the bullet to travel the 5 feet from the gun to your body. If you could do stuff that fast, then you could get a lot of experiences into a small amount of time without the help of a machine. It’s just easier to tell the story with experience machines.
Dale M,
The sorts of cases you’re talking about are possible too, but I have in mind a case where it seems, *at the time the experience is happening*, that it’s going a lot slower than it really is. Isn’t that possible too?
So, Ben, does Bob get as much pleasure in the short time as Alma gets in the longer time? As I understand the story, I’d say yes.
Ben B.,
Do you die when you get off (or just before you get off) of machine A, too?
Jamie: I am inclined to say he does. So I think Benjamin Bagley’s idea is probably right. But the other way to go would be to say (contrary to Pablo) that we are fallible judges of how much pleasure we’re getting, even while we’re getting it. Suppose that were true; then we could wonder whether what matters is how much pleasure we’re really getting, or how much it seems like we’re getting.
John T: yes.
Ben,
This might just be a housekeeping matter, but shouldn’t it matter that, other things remotely equal, getting the pleasure all in one day seems to be a much surer thing than spreading it out over 120 years?
After all, the A-machine might break down after, say 30 years, leaving you short on pleasure. The B-machine, by contrast, need only operate unproblematically for 1 day.
That makes me think that B is the obvious choice, in terms of expected value, for hedonists.
Maybe you could offset this by stipulating that the B-machine is extraordinarily unreliable, thereby eliminating the disparity?
John,
I’d rather just stipulate that both machines are 100% reliable. The example is supposed to test intuitions about well-being, not rationality.
Um, you asked, “Which machine should you choose?”
Oh yeah, sorry. I guess it would have been better if I’d asked which machine provides the better life.
Ben,
“The sorts of cases you’re talking about are possible too, but I have in mind a case where it seems, *at the time the experience is happening*, that it’s going a lot slower than it really is. Isn’t that possible too?”
Honest answer: I have no idea. A little Googling turned up this article, which is interesting even if it doesn’t provide definitive evidence that what you’re suggesting doesn’t really happen.
http://www.livescience.com/health/071211-time-slow.html
Ben,
I can explicate what I tried to say earlier only roughly, as follows. The duration of an experiential episode equals the sum of the distinct moments it contains. Imagine a series of person-stages, all belonging to the person who is said to experience the episode, each corresponding to one distinct moment of that episode. Suppose it seems to each of the person-stages that their respective occurrent experiences last a moment. There is a sense in which it then seems to the person who comprises those person-stages that the episode lasted its duration. When I said that if some experience seems to last a certain amount of time then it lasts that amount of time, this is the sense I had in mind.
Perhaps this is getting unnecessarily complicated. I can simplify my line of argument by trying to get you answer these questions. Do you think that there is a fact of the matter about the duration of phenomenal experiences? Or do you think that all we can say about experiential episodes is that they are judged retrospectively to be of a certain duration? If you believe there is a fact of the matter, is this a physical or a phenomenal fact?
Oh yeah, sorry. I guess it would have been better if I’d asked which machine provides the better life.
Notice the Nozickean parallel. Nozick asked his readers whether we would plug in to the machine, but in fact wanted to test whether plugging in would be best for us. (“Would you plug in? What else can matter to us, other than how our lives feel from the inside?”)
Dale M: very interesting! I guess I’d have to fall back on the conceptual possibility of Machine B.
Pablo: Yeah, but I have a better excuse than Nozick. I was writing a blog post, not a book!
I will have to give your questions a bit of thought.
I’ve a follow-up post here. Excerpt: “It’s one thing to write a story which says “100 years passed”, and quite another to fill out the details for 100 years’ worth of fictional events. So, if we think of Machine A as taking a long time to impart rich informational content (lots of pleasure), and Machine C as taking a very short time to impart a very thin representation (very little pleasure), the question whether Machine B imparts a lot or only a little pleasure comes down to the richness of the representations it implants…”
Richard: thanks, I’ll check it out.
Pablo: Yes, I think there’s a fact of the matter about how long experiences last. In Machine B, the experiences are supposed to last 1 day. As for whether it is a physical or phenomenological fact: I don’t know, because I don’t know whether phenomenological facts are a species of physical fact. (Nor am I very confident about being able to say what exactly a physical fact is.) But even if phenomenological facts aren’t physical facts, I think the experiences last only a day. I don’t know if that is helpful or not.
I have some worries about what you say about adding up moments to get durations. Won’t there be the same number of moments in a short duration as in a long one? Or are moments bigger than point-sized?
Ben,
In Machine B, the experiences are supposed to last 1 day.
I hadn’t realized Machine B was by assumption supposed to induce experiences that lasted one day; I thought the description of the situation was neutral with respect to this question. As Richard puts it in his reformulation, “Machine B (allegedly) gives you all the same experiences, feeling exactly the same from the inside, but packed into just a single real-time day.” (In your original description of the case, you said that “the experiences are crammed into a much shorter period of time: just one day. […] At that time it seems to you as if you have lived 120 highly pleasant years.” This “textual evidence” is compatible with both your later reformulation and my original interpretation.) In any case, the interesting philosophical question seems to me to be whether the experience induced by this machine is more accurately described as an experience that lasts 120 years or as one that lasts merely a day.
To answer this question, we may consider a different thought experiment. Imagine a possible world containing nothing but two unembodied minds, Mind A and Mind B, each of whom undergoes only an experience that is qualitatively identical to those induced by Machine A and Machine B respectively. Would you say that Mind A and Mind B undergo experiences of the same duration? And, in the light of the answer to this question, what do you think hedonists would say about the value of the experiences induced by Machine A and Machine B? It seems to me quite clear that, if the answer to the first question is Yes, then hedonists would say that both machines produce experiences of equal value. This seems particularly clear because, being a hedonist myself, this is what I would say.
I have some worries about what you say about adding up moments to get durations. Won’t there be the same number of moments in a short duration as in a long one? Or are moments bigger than point-sized?
I was relying tacitly on the distinction between instants and moments; the former are point-sized, while the latter have some temporal duration. So, the answer to your questions is No and Yes respectively.
Thanks Pablo, that is helpful. I am inclined to agree with you about the values of the experiences in the disembodied mind case. I’ll have to give the duration question more thought, I think.
I think machine A is the superior machine, again with most people’s intuitions, the duration of the experience may appear to be the same, but I’m not convinced that it could actually be the same.
If they really were the same, then shouldn’t both hedonists pick machine B? Jump in, experience 100 years of pleasure, then jump out before you die, and repeat. But I think that defeats the point of the thought experiment. 🙂