Is the Simulation Argument an Improvement on the Dream Argument?

by Tim Sommers

“Suddenly he woke up and there he was, solid and unmistakable Zhuang Zhou. But he didn’t know if he was Zhuang Zhou who had dreamt he was a butterfly, or a butterfly dreaming that he was Zhuang Zhou.” — Zhuangzi (translation by Burton Watson)

“We are almost certainly living in a computer simulation.” – Nick Bostrom*

Is the hypothesis that we live in a computer simulation an improvement, in some way, on the classic skeptical argument that life is but a dream? The dream argument seems to show that life could be a dream. Some claim that the simulation argument shows that not only is it possible that we live in a computer simulation, but that we almost certainly do live in a simulation.

I’ll argue that the simulation argument does not make it any more likely than the dream argument does that this is not reality. Furthermore, the simulation argument might even be worse (as a skeptical argument) in one way. If I am dreaming, there is not just another world, but, in some sense, another me, out there beyond the dream. But if I am a being that only exists in a simulation, it follows that there is no other me out there – and challenges the very idea that this scenario is really “skeptical,” in the same way, as the dream argument.

From Zhuangzi to Descartes to The Matrix, people have worried and wondered over global epistemological skeptical scenarios like these. Let’s call them GESSes.

They are global because they cover all knowledge from our senses, they are epistemological because they raise the question of what we can know, skeptical because they answer, ‘we can’t know anything,’ and scenarios in the sense that they offer a story about why our senses systemically fail to make contact with reality.

So, how do you know whether we are currently being deceived about everything around us by an evil demon or dreaming it all, as Descartes says, or if we live in some kind of simulation?

Though some philosophers have tried to argue that dreams are sufficiently distinguishable from real life to defeat the argument that life is but a dream,  there haven’t been many attempts to test whether a particular GESS holds or how likely it is; i.e., whether there is actual evidence that life really is a dream or a computer simulation. (There’s a good reason for that, which we will get to eventually.) But there have been a few purportedly scientific attempts to address the simulation  argument. And there is a popular meme, you’ve probably seen, going around right now, photos or stories of supposedly real life “glitches in the matrix” or “glitches in the simulation.” That is, of purported cases when the computer simulation we live in breaks down: déjà vu, inexplicable duplicates of people appearing, or everyday objects appearing pixilated. The phrase “glitch in the matrix” comes from the film “The Matrix,” of course, where, as you probably know, (spoiler!) the world is ruled by AIs that keep humans as batteries, confined to sleeping pods where they are distracted by a mass computer simulation of the world of the as it was in the 1990s – an era the AIs correctly identify as the greatest era in human history.

As far I can tell, meme makers don’t really distinguish between the matrix GESS, which we might call the “brain in the vat” scenario, and the “simulation” GESS. But there is this a difference. In the brain in the vat GESS you really are a brain, or a body, or some physical thing, plugged into some kind of simulation. In the simulation hypothesis you are inside the simulation. You are part of the simulation. You, yourself, are a simulation.

So, the dream GESS, probably the foundational GESS given we have always been dreamers, is presumably a matrix/brain in the vat GESS and not a simulation GESS. Why? Because the dream that you are experiencing now as your life must be the dream of someone or something somewhere. But in the simulation, you could be in, you don’t necessarily exist outside the simulation. In fact, I would say you necessarily don’t exist outside the simulation.

You could object that, in the Bostrom version, for example, the hypothesis is that we are part of an “ancestor simulation.” That would mean that you are the simulation of a particular person that existed and was not a simulation. But this doesn’t collapse the simulation hypothesis back into a dream GESS, because, even to the extent that you are a simulation of some particular person, you are not that person. If I use a teleporter to make an exact copy of you without destroying the original, the copy is not you, right? (Even if it is just as good a way of going on as the regular way. See, my “Would it be Better, If there were More of You?”) Even if you could be uploaded into computer, the version of you in the computer is not you, right, but a digital copy? So, an ancestor simulation does not make you into the simulated ancestor, you exist apart in a fully virtual digital world.

Why think the simulation GESS is an improvement on the dream GESS, then?

The argument is that there are likely many simulations, but only one baseline reality.  Bostrom says that “the credence one assigns to being in a simulation gradually approach[es] the limiting case of complete certainty” when we contemplate the possible proliferation of simulations (though for reasons I don’t have time to get into here, his official view in that paper is that there is a 33.33…% chance we live in a simulation). To repeat the key move: there’s only one reality, but lots of simulations.

The first response should be to say, sure, but aren’t there are also many more dreams than dreamers? There are many nights, after all. And there’s no reason to think that whoever makes all these simulations doesn’t also dream. Why think there are not just as many dreams as there are simulations? In which case, the probability-game favors neither view.

One response would be to say that, as far we can tell, we spend more of our lives not dreaming than dreaming. But this takes me to the second, and for me, decisive response.

All of these considerations about how many dreams there are, and how many simulations there could be, are extrinsic to these skeptical scenarios. GESSes show us that we may not know certain things. I can’t make it more plausible that the GESS actually holds by introducing a lot of facts that, about the GESS or outside world that, by hypothesis, I don’t know – and worse don’t know anything about the probabilities thereof. In other words, if this is a computer simulation and we don’t know anything about the real world, even if we can make sense of the question, we have no information to determine how likely it was that we would exist in a computer simulation. And if this isn’t a computer simulation, we have no idea if such a thing is even possible or, if it is, how likely it was that we might have been in one.

There’s a deeper question about whether probabilities ever make sense in metaphysical contexts. I can wonder whether only physical things really exist (whether all concrete particulars are physical, to use the physicalist slogan) or whether there are mental properties that don’t reduce to physical properties, but it’s nonsense to say that there there are two possibilities and, therefore, a 50% chance that either is true. Either physicalism is true or it isn’t. There are no odds that it is true. We either live in a dream or a simulation or we don’t. That doesn’t make the odds 50/50, nor does it open up the possibility of arguing (as Bostrom does) that the odds are greater (or worse) than 50/50.

If I grant that we currently live in a simulation, in other words, I don’t have any way of calculating the probability of what’s going on in the “real” world I don’t live in; i.e., whether it is like, or totally unlike, the world in the simulation, whether trillions of other simulations are running there, or that this is the last simulation in the universe abandoned on a dead world and running out the dregs of its battery power, or this is a simulation so unlike simulations as we know them as to be utterly incomprehensible to humans how it fits into the larger world. “Incomprehensible,” by the way, might actually cover every case of the claim that a wholly accurate simulation phenomenally indistinguishable from our world could be created by running a Turing machine. I know we all pretend to understand that possibility. But do we? Really? Dreaming is a real thing. But you must admit that simulations indistinguishable from reality are speculative, at best, no matter how good the graphics are of the latest Assassin’s Creed.

Based on these GESSes, I can only say you may be dreaming or in a simulation. Any further facts I introduce are hostage to the original skeptical scenario. It’s as if I offered to buy something on credit, and then put down a deposit – also on credit – to prove I was going to pay the first debt.

It doesn’t help to introduce the idea of levels of simulations or dreams within dreams or simulations, either. A dream within a dream is just a dream. A simulation within a simulation is just a simulation. If I say you may be dreaming or in a simulation, I can’t add that it is more likely that you are simulated or dreaming than not, because, in the dream or simulation you are now in, certain facts about dreams and simulations hold true.

Finally, we come back to the way in which, I think, the simulation is actually a weaker skeptical scenario than the dream argument.

David Chalmers argued in his recent book that simulated or virtual worlds can be just as good as the real world. The argument, it seems to me, is a bit of a cheat. I mean, if you stipulate that the relevant virtual world is “just as good as reality,” aren’t you just begging the question when you announce later on that it, yes, it is? On the other hand, maybe Chalmers is on to something. If my only existence is an existence in a virtual reality, and I have no connection to anything outside the simulation, am I really wrong in most of my knowledge about how my world works – just because I don’t know that this is not the only world?

Compare the possibility raised by string theorists that “The three-dimensional world of ordinary experience – the universe filled with galaxies, stars, planets, houses, boulders, and people – is a hologram, [a three-dimensional] image of a reality coded on a distant two-dimensional surface.” Is that a GESS for the third-dimension? Or should we say that, for non-physicists especially, whether the third dimension is holographic has nothing to do with claims like the one that last night I looked up and down and all around for you? If epistemological skepticism simply amounts to the claim that the world is very different from the way we perceive it to be, well, I guess, physics proved that a long time ago.

*I recognize that Nick Bostrom is a problematic figure. But, as far as I can tell, he was the first to argue that the simulation argument is an improvement on the dream argument. I don’t want to discuss him, just that.