Learning from Experience Machines

No, fully convinced that the experience will be veridical (non-illusory).

You're missing the point of the thought experiment.

But if the real experience and the illusory version are both “fully convincing,” but the illusory one offers a more valuable (to you) lifelong experience, then why wouldn’t you step in?

It depends on "fully convincing to whom/when". If you are deciding *now* whether to step into an illusion machine or not, then you know *now* that it is an illusion and so it does not convince you *now* (at the moment you make your choice). And that is the person that needs to be convinced: the person making the decision, *at the moment* that he makes the decision.

A machine that is "fully convincing" in the sense that, once you step in it, it will erase your memory that it is an illusion, is still not fully convincing to you *now*, as you make the choice whether to step in.

Scott,

If the illusory experience never ended, you're correct, cognitive dissonance could be avoided. But if the real experience and the illusory version are both "fully convincing," but the illusory one offers a more valuable (to you) lifelong experience, then why wouldn't you step in?

Fully convinced of what, now - the assurance of being fully convinced of a better than real experiences?

No, fully convinced that the experience will be veridical (non-illusory).

I guess I don’t understand Nozick’s point, or yours. If you’re “fully convinced” of the experience, why wouldn’t it seem real to you?

RKN, it would seem real. But it would not be real. I'm assuming there's a meaningful distinction between the two because I believe there is one. Nozick's point was, I believe, even if we are given the choice between something that seems real and is and something that seems real and is not, we should choose the one that actually is real.

And I agree.

I disagree RKN, and though I don't have the book onhand, I think Nozick addressed that objection. We could, conceivably, create an experience machine that would never end, and as such there is no need to worry about the cognitive dissonance that results from disconnection. We could even, conceivably, erase some part of the individual's memory before he stepped into the machine, so that he would have no idea that the experience machine was false.

Even with those two possibilities, I wouldn't step into the machine. I suspect many others wouldn't as well.

Because it wouldn't be real, and I prize being in touch with reality for its own sake.

That was Nozick's point.

Scott,

I guess I don't understand Nozick's point, or yours. If you're "fully convinced" of the experience, why wouldn't it *seem* real to you?

Everyone get a good long gaze at Scheule's navel, there?:wall:

But even if it narrows completely, nevertheless we may retain a lingering awareness that the experiences are not real. And this will often be enough to drive us to seek out the genuine article rather than the simulacrum. However, as time passes and we become accustomed to the simulacra, we may come to regard the simulacra as merely a different sort of real, rather than as not real. For, how do we know what is real? How do we tell reality apart from illusion?

Excellent questions. The problem doesn't seem to me to turn on what "real" means, so much as it does on what "convincing" means. The thought experiment posited "a fully convincing illusion of an experience." If you write and publish a book for real, then presumably you come away fully convinced you have. Being convinced you have means you believe it is true you have -- a state of mind. So if the machine is "fully convincing," that you have written and published a book, then you necessarily believe it is true you have written and published a book.

The reason to prefer the real and avoid the simulacrum is to avoid the severe bouts of cognitive dissonance that would result when the illusory experience ended and you were unable to read the book you're "convinced" you wrote, or hug the children you're "convinced" you have.

You’re missing the point of the thought experiment.

No, I'm not. I'm explaining why Scott and Nozick choose what they choose. Breaking it down, analyzing it in detail. You are the one who evidently does not understand what Scott and Nozick are getting at with the thought experiment, since you don't understand why they make the choice they make.

I think that's an insult, but I don't get it.

Because you are making the choice now, before you step in, and you are not now fully convinced.

Fully convinced of what, *now* - the assurance of being fully convinced of a better than real experiences?

The experience machine would not suffice because things do not reduce to our experience of them. And our desires, which concern things, therefore also do not reduce to desires for experience.

Now, let's be careful. Part of the experience of something is the belief in it. So in order for VR experience to be fully like real experience, then we would need to be unaware that it was VR. And if we were unaware, we might be perfectly content with VR. But of course, we would be content only because we were fooled. And, considering the matter in the abstract, we would not want to be fooled.

Speaking from the point of view of evolution, which shaped us, our experiences are mere markers of the really important events, which concern real stuff like whether we really reproduce. Evolution has made us pursue certain experiences only because, in pursuing them, we pursue the really important events and outcomes.

We've learned to have many of those same experiences independently of the important outcomes that they normally accompany. A VR world would be a culmination of this trend. For the moment, our artificially induced experiences are usually not as rich as the genuinely real experiences, and so even from an experiential point of view there is a difference between mere experience and reality. However, sometimes that difference can be narrowed well enough that we lose the ability to discern reality from illusion. In the future it may be narrowed further.

But even if it narrows completely, nevertheless we may retain a lingering awareness that the experiences are not real. And this will often be enough to drive us to seek out the genuine article rather than the simulacrum. However, as time passes and we become accustomed to the simulacra, we may come to regard the simulacra as merely a different sort of real, rather than as not real. For, how do we know what is real? How do we tell reality apart from illusion? One of the main methods is historical priority: what came first is the original, and what came second is the copy. The Sony robot dog is the copy and the real dog is the original, and we know this because the real dogs were here first. But imagine growing up in a world populated all along (as far as you have experienced) by flesh creatures and by robotic creatures. Which one, then, is real?

I offer as an example of this, homosexual marriage. To people who have only experienced heterosexual relationships, it is likely that homosexual marriage will seem a simulacrum of heterosexual marriage, and not a very good one at that. But to those of us with heavy experience of homosexual relationships, homosexual marriage is more likely to seem to be another genuine variant of marriage.

They say that higher animals, as sentient beings, have some of the substance human emotion. Your piece makes me wonder how the sexually aroused dogs behind the house felt when I broke them apart with the garden hose.
But seriously "We want experiences, fitting ones, of profound connection with others, of deep understanding of natural phenomena, of love, of being profoundly moved by music or tragedy, or doing something new and innovative, experiences very different from the bounce and rosiness of happy moments. What we want, in short, is a life and a self that happiness is a fitting response to—and then to give it that response."--- Comments to David Friedman’s post 7/14 2006. Now I must retire to the back yard, my pit bull is barking and wants to play.

My question is: if I step into the machine and my (slightly better) experiences from that point forward leave me “fully convinced” they happened, why not do it?

Because you are making the choice now, *before* you step in, and you are not *now* fully convinced. You will be in the future, but it is *now* that you are making the decision, and your decision will be based on your *current* beliefs.

Saying I prefer so-called “real” experiences to the simulacrum presumes one would not be “fully convinced” by the simulacrum.

No, it does not. What you prefer (present tense) now, the choice that you will make now, depends on what you *are* convinced of now, not on what you *would be* convinced of later.

But a premise of the experiment is that you will be.

Indeed. Later, you will be. But now is when you make the decision.

Nozick’s point was, I believe, even if we are given the choice between something that seems real and is and something that seems real and is not, we should choose the one that actually is real.

If that's all he concluded I'm even more unimpressed. You don't need a thought experiment for that. The question remains: *why* should you choose the real?

All normal human beings have seemings. A seeming is the sum of our sensory inputs. We use our seemings to draw conclusions about what is real. Usually our seemings lead to correct conclusions, but sometimes not, because senses can be wrong. When you reach a state of mind of being "fully convinced" an experience is real, it is because your seemings "network" has validated it as real. If a machine promises experiences will be "fully convincing," that implies it can "program" (somehow) your seemings network directly, bypassing the sensory inputs. In such a case a human mind would be unable distinquish an experience as having arisen from reality or a simulacrum. And, I argue, unable to form a preference for one or the other.

*But*, if I'm promised the experience in the machine will be overall *better* (as the premise did) than what reality will deliver me, then it "seems" to me I might want to step into that machine. Especially with the added bonus that it would first erase my memory.

A machine that is “fully convincing” in the sense that, once you step in it, it will erase your memory that it is an illusion, is still not fully convincing to you now, as you make the choice whether to step in.

Erasing my memory of the past isn't the lure to step in, it's the promise to experience a slightly better life. If this entails erasing my memory of the past, all the better.

My question is: if I step into the machine and my (slightly better) experiences from that point forward leave me "fully convinced" they happened, why not do it? Saying I prefer so-called "real" experiences to the simulacrum presumes one would not be "fully convinced" by the simulacrum. But a premise of the experiment is that you will be.

I must confess that I've never quite understood Nozick's (or I guess Scott's or Constant's) intuition in this case. Why should the world of protons and electrons that I currently inhabit be somehow innately (and quite mysteriously, I might add) preferred to an alternative world of protons and electrons that I could inhabit? What is it about this sort of world that makes it inherently better to the other?

Let me offer a different thought experiment. Suppose that you were nearing the end of your life and I told you that I could download your mind into a virtual world where you could continue to have all sorts of experiences. All those experiences would take place within the virtual world, but those virtual experiences would be no different in terms of how they felt (i.e., they would be fully convincing and would seem completely veridical).

Now if Scott is right, you ought to prefer the "real" experience of suffering and then dying to the "fake" experience that I offer you in the downloaded virtual world. I submit, though, that anyone who chooses to die in the "real" world rather than continue to live in the "fake" one is either (a) caught in the grips of a (non-hedonist) ideology, or (b) not completely understanding that the experience really will be fully convincing. That means that at no point, ever will you come to know that the whole thing is "faked" (whatever that means). I think that (b) is really the more common response to Nozick's own version, as well. Our objection lies in our failure to really understand and appreciate what it means to have a fully convincing virtual reality.

Indeed, why not think of the experiment as simply offering a different way of having experiences in the world? What is it that I value about life if not the experiences that I have? Why does it matter whether those experiences are of other organic beings or are of programs that are really indistinguishable from other organic beings? To say that one likes something because it is "real" is to assume that only the natural organic world is actually real. I can't see any good reasons for thinking that to be true. Or at least I can't see any non-mystical, unromanticized reasons for thinking that to be true.

I must confess that I’ve never quite understood Nozick’s (or I guess Scott’s or Constant’s) intuition in this case. Why should the world of protons and electrons that I currently inhabit be somehow innately (and quite mysteriously, I might add) preferred to an alternative world of protons and electrons that I could inhabit? What is it about this sort of world that makes it inherently better to the other?

No, not my intuition. I wrote: "However, as time passes and we become accustomed to the simulacra, we may come to regard the simulacra as merely a different sort of real, rather than as not real." (Nor do I say this is necessarily the case - I am pointing out that our views *can* evolve this way.)

To explain further, *if* you want to have children, then you want to *have children*, which is not the same thing as seeming to have children. Therefore if you want to have children and someone offers you the opportunity to have the delusion that you have children, without actually having children, then you will refuse, so long as your desire is *to have children*.

Look at it this way. One of the reasons people want to have children is so that some part of them will survive their death. Therefore, to the extent that that is a motivator, then what people want is something that lies outside their own experience because it survives their death. Therefore what they want *cannot* be reduced to mere experiences and cannot be fulfilled merely by them being deluded into believing that they have children.

However, as I pointed out, our views can evolve. We may come to consider robots as perfectly good substitutes for biological children. We may come to consider them as children of a different type, and not merely as pretend children.

Indeed, why not think of the experiment as simply offering a different way of having experiences in the world? What is it that I value about life if not the experiences that I have?

If you want to have children because you want some part of you to survive your personal death, then you value something other than the experiences yu have.

If, dying, you hope that your children live happy lives, then your hope concerns something that lies outside of the experiences that you can have, and so you value something about life other than the experiences you have.

If you devote years of your life and a large fraction of your income to give your children the best chance at a happy future, then you value something about life other than the experiences you have.

Our behavior is driven by our genes. We are programmed to want certain things, and the ultimate things we are programmed to want concerns the survival and replication of the genes.

But the survival and replication of the genes is a reality that lies partly outside of our own experience. Therefore it is not surprising that we have desires that go beyond what lies within our own experience.

Now, you can of course ask, "isn't it irrational for us to let ourselves be slaves to these genes," but essentially you are proposing that we *choose* a different basis for our behavior, an experiential basis, a basis that makes our experiences the ultimate goal of our actions. But this is merely a proposal rather than an observation. One could as easily make any number of proposals. What is so great about yours?

The questions that the thought experiment raises are another matter; the fact is that many of our intuitions prefer a choice that neither Joe nor RKN can grasp the reason behind. Now perhaps this indicates Nozick and my intuitions are irrational---we really should have no preference for reality over the experience machine. Or, alternatively, it indicates that there is a significant answer to the question of "Why prefer reality?" (perhaps not uncovered yet).

I think the latter is true.

As to Joe's objection, I will simply present an interesting tension in his own belief structure that I think---when resolved---may be illuminating for him, if not for the rest of us.

Joe,

To say that one likes something because it is "real" is to assume that only the natural organic world is actually real. I can't see any good reasons for thinking that to be true.

Putting the choice in terms of organic vs. non-organic is misstating or missing the point (we could replace the experience machine with some sort of holodeck that used organic matter in creating its apparitions), but I don't want to concentrate on that. I'd like instead to point out that your are committing what you would call a definitional dodge.

Earlier, on your blog, you described rational egoism (or something like that) as being an ideology with laughably bad defenses. (You also wrongly attributed the position to Ayn Rand, as Matt commented.) You pointed out that many proponents of such a view defended it by saying that all actions were effectively selfish, in that they are taken to satisfy a want of the actor.

This, you said, is a definitional dodge, because people are using selfish in an atypical way. (More precisely, I imagine what people are doing is stretching a verbal definition [of the word "selfish"] until it includes actions that no longer fit peoples' intuitions.)

Now maybe it is, maybe it isn't. But what is interesting is that you, attempting to call "virtual reality" "real" are performing an identical maneuver. Consider the similarity of the succeeding two paragraphs.

All actions are selfish. What is a selfish act but something taken to satisfy your own wants? And if you want to help somebody else, and you act to do so, aren't you simply satisfying a want of yours, the want to help someone? Of course you are, as you would be doing any time you acted. Therefore all purposeful acts are selfish.

All things are real. Every dream, artifical world, hallucination, is ultimately reducible to the electrons and protons in your brain that perceive the experience. And are not protons and electrons real? Of course, and thus anything made of them must be. Therefore all things are real.

Note how each deconstruction has an identical significance: both imply that the words they dissect, "real" and "selfish," are meaningless, and that people who use the terms are being irrational (or, in your terms, mystical and romanticized).

But you buy one criticism and (soundly?) reject the other, and therein lies the tension. Now there's nothing prima facie wrong with your belief structure---our beliefs are mobile, and tension is unavoidable with so many moving parts. I admit there are a great many tensions in what I believe.

There are four obvious possibilities related to this:

(a). The realness of something is a meaningless concept, but selfish- and selflessness are indeed meaningful.
(b). Realness is meaningful, whereas selfish- and selflessness are meaningless.
(c). Both realness and selfish/selflessness are meaningful.
(d). Neither realness nor selfish/selflessness is meaningful.

Now, I think (c) is correct, and it would seem that you think (a) is correct. And whether we are right and if so, which one us is right, would be a long discussion that I have no intention of getting into here. Nonetheless, simply pointing out the situation and the underlying tension is, I believe, illuminating in itself, because I believe that ultimately, the reason that I know "realness" is a meaningful concept is the same reason you know "selfishness" is a meaningful concept, and that (d) is false because of it. Perhaps it is mystical and romantic to think that some situations are more real than others, but it seems to me that those labels must be branded just as surely on the view that some actions are more selfish than others.

In short, the Nozickian thought experiment brings to light an intuition that may be corresponding to an underlying truth.

As to your hypothetical---if what you're describing is simply an experience machine that we had the choice of entering late in life, my answer is no, I wouldn't enter it, and being a science-fiction buff I doubt the reason is that I simply can't conceive of what a convincing world would be. Nor do I seem to be in the grips of any particular ideology (I am, by most accounts, a rather poor libertarian). But others may judge.

At any rate, with that, I've defended myself to my satisfaction, if not others'. I'm sure the experience machine experiment has more thoughtful proponents and critics than me: feel free to search them out. I'm going to the classical music forums, where the discussion is not quite so challenging.

And we’re still waiting for yours, should you care to offer one.

I don't claim to have an answer. In fact, in the original post I indicated that the reason for choosing reality over simulated experience is mysterious. Nonetheless, my intuition tracks a real truth.

It's pertinent to note that eventually any reason I give for the preference (or against it) will bottom out in some base intuition, i.e. it doesn't seem right to do so and so, or live in such a manner. The same thing happens with any should claim. But that's the nature of the is/ought divide.

In analyzing this situation, you have to remember that in real life persons always try to alter reality. Say you are entering a race. Postulate that your life will be improved if you win the race.
Instead of entering the Experience Machine you try to win the old fashioned way. How far would you go before preferring to enter the Experience Machine?
1.) Vigorous training and good nutrition.
2.) Hire world’s best coach, and buy best equipment that no other opponent can afford.
3.) Wear a lucky charm such as a rabbit foot.
4.) Take Epo, amphetamines and anabolic steroids.
5.) Bribe the judges. Put nausea inducing drug in your opponent’s food. Hire thugs to beat up your opponents.

Then enter the Experience Machine and before entering it tell it what virtual acts you will allow it to commit in order that you may win in the machine’s world. Would it make any difference to you whether the machine didn’t let you know that it had poisoned the characters in the virtual world so you could win?
I think the answers will show how tightly we are linked to other people. Since there are no people in the machine world, nothing you do makes any difference, while in the real world you have that complex continuum of selfishness vs. fairness and Sneakiness vs.honesty

The questions that the thought experiment raises are another matter; the fact is that many of our intuitions prefer a choice that neither Joe nor RKN can grasp the reason behind.

The thought experiment has nothing whatsoever to do with intuitions, or "grasping reasons behind them."

Which, btw, is a truly curious statement, since if anything we normally think of intuition informing reason, not the other way around.

Now perhaps this indicates Nozick and my intuitions are irrational—we really should have no preference for reality over the experience machine.

This is not what this experiment tests. If your future experiences inside the machine will be (value-wise) the same as those reality would deliver you, obviously there is no incentive to step into the machine. That's why in this experiment - the one *you* linked to in the blog post - it was given that a *better* (value-wise) experience than what reality would deliver you awaits you inside the machine. It could do this because it was also a given that the machine had somehow figured out what your future in reality would be. Combine these *givens* with the promise of a "fully convincing" set of future experiences (yet another given), which I'm sure was intended to mean every bit as real to your seemings as those in reality, and this arguably changes one's preference. If not, *WHY* not.

Or, alternatively, it indicates that there is a significant answer to the question of “Why prefer reality?”

And we're still waiting for yours, should you care to offer one.

Scott,

Thanks for the thoughtful response. For what it's worth (if you're not too busy with the classical music forum), I've posted a response here.