Elon Musk Is Wrong. We Aren't Living in a Simulation
A simulated apple can't feed anybody, and other reality checks by a philosopher and a cognitive scientist.
Illustrations by 30000fps
Recently Elon Musk made internet headlines by claiming that the probability we live in "base reality" is one in billions. Instead, we are much more likely to be living in a historical ancestor simulation created by an advanced future civilization some 10,000 years from now.
"The strongest argument for us being in a simulation, probably being in a simulation is the following: 40 years ago, we had pong, two rectangles and a dot," the SpaceX and Tesla founder said. "That is what games were. Now 40 years later we have photorealistic 3D simulations with millions of people playing simultaneously and it's getting better every year. And soon we'll have virtual reality, augmented reality, if you assume any rate of improvement at all, the games will become indistinguishable from reality." He continued,
"There's a one in billions chance [we're in] base reality... I think it's one in billions. We should hope that's true because otherwise if civilization stops advancing, that could be due to some calamitous event that erases civilization, so maybe we should be hopeful this is a simulation. Otherwise, we will create simulations that are indistinguishable from reality or civilization will cease to exist. Those are the two options."
It is probably not Musk's most original idea. The notion that we live in a simulation, sometimes called the simulated universe hypothesis, was popularized in the 90s by the Wachowskis in the The Matrix, but the idea has been around for years. In an episode of Doctor Who from 1976 called "The Deadly Assassin," a community of individuals live a simulated life inside a machine called—guess what—the Matrix!
In fact, Musk is elaborating a technologically updated version of one of the oldest philosophical puzzles—namely, is the world we live in nothing but a dream? Basically, it is the Kantian idea that experience is a phenomenon, while the real world—the thing in itself—is hidden to us. Or, if one wants to go back to the Greeks (who doesn't?), it is the allegory of the cave: What we believe is the real world is nothing but the distorted shadows cast on the cave's wall by simulacra (statues and silhouettes) of actual objects.
Thinkers kickstarted philosophy with the contrast between appearance and reality—i.e., what people took to be real is only a simulacrum of the true world (simulacrum and simulation). And since then, the main point of the philosophers' job description was, of course, to figure out what the true world was, which is what Elon Musk calls "base reality."
Today, we have computers, software and mathematical models that allow us to use a much more scientifically oriented jargon, but the conceptual outcome is not that different. Yet, the notion might remain deeply flawed. The point raised by Musk is not really whether it is conceivable that a future society will invest in detailed simulations, but if the world we live in can be a simulation. Is a simulation the kind of thing that might be, given sufficient computational powers and details, mistaken for a world? Or are simulations and worlds fundamentally different things?
Simulations Are Things In The World
The key question is what are simulations made of? Or, if you are more poetically inclined, what is the stuff that dreams are made of? Simulations are things that we use to talk or to think about other things. In this respect, they do not step out of Musk's base reality. They are still base reality. They are made of the same stuff everything else is made of.
For instance, a 10-inch neoprene model of Mount Everest is still an object, albeit an object that is used to refer to a much bigger object. A flight simulator is a physical thing used to refer to real planes. A dynamic simulation on a computer of the galaxy is yet another object made of rather complex networks of electronic gates and devices cleverly connected. It is a dynamic object we use to refer to another object. But nowhere do we meet a pure simulation that is not an object.
The idea that simulations are a sort of immaterial entity that are, despite being dependent on their physical substrate, nonetheless different, is a leftover of the aforementioned belief in a higher—and possibly better—reality. It's a belief that we have no reason to take seriously. The notion that we may mistake a simulation of the world for the world is both conceptually and empirically flawed.
Conceptually, it is a self-defeating notion—something that if taken to be truth, negates itself. In fact, if, say, simulated water might be a meaningful notion, what would it be made of? It could not be made of real stuff, because if it was, it would no longer be simulated water. However, neither could it be made of simulated stuff, because—that's the point of being a simulation—there is no such thing as simulated stuff. All we know is physical. All we know belongs, once again, to base reality. Either way, simulated water cannot exist.
Empirically, increasing computational power will not necessarily transform the water of computer games into the wine of a full-fledged simulated world. Making bigger bows and stronger arrows will never lead to an H-Bomb. Sometimes there are conceptual gaps that cannot be bridged by incremental improvements. Living in a simulation is not like building a 1-mile-high tower, which is challenging but possible, but rather like having a planet with a certain mass and no gravity. No amount of technological progress will achieve the latter, no matter what.
Moreover, Musk's confidence in the development of technology—that massive increases in computational power will transmogrify existing videogames into a real simulated world—is based on the confusion between the ideal notion of simulation, which does not really exist, and the actual thing a simulation is. The ideal notion of simulation is, in turn, based on the notion that there are disembodied minds or a higher level of reality over and above basic reality. This is highly questionable.
The apple that looks so convincing to Musk on a VR computer screen would be utterly disappointing for a butterfly looking for a home or for a Robin looking for a worm
In a nutshell, what could this simulation be made of? If the reality we see were a simulation, we should assume that the simulator is made of altogether different stuff that, by definition, we could not even conceive (it should be made of something completely different from everything we meet in our world). While the notion that there is a base reality and additional levels of reality is both appealing and enthralling, we have evidence of only one level of reality. The world we live in is just made of objects.
After all, this is confirmed by science itself, whose equations describe the flow and interactions of one kind of stuff—matter and energy according to relativity theory and quantum mechanics. No additional stuff appears in the scientific description of reality. We have, say, planets, and computers we use to anticipate what planets will do, but we do not have fancy planets inside computers. Using computers, which are objects, to predict what planets will do is what we call a simulation. There are no little fancy simulated planets anywhere in the computer. There are electronic gates we use to talk about planets. Taking the existence of a simulated level of reality too seriously is akin to believing in the existence of "the nobility" as something over and above the power of the rest of society.
Musk's argument is further plagued by two more hidden assumptions. First, the notion that games are becoming indistinguishable from reality. They are indistinguishable from reality only because we grant them all kinds of allowances. But a simulated apple cannot feed anybody. No matter how many pixels depict it. Conventionally, when we compare a real apple with a simulated apple, we leave aside all properties that cannot be simulated. In the 80s, the first version of the Microsoft Flight Simulator looked terrifically convincing to legions of players. It no longer does. The apple that looks so convincing to Musk on a VR computer screen would be utterly disappointing for a butterfly looking for a home, or for a Robin looking for a worm.
Yes, Musk might rebut that in the distant future the "posthumans" may add to their simulation more than just colors on a computer screen. They might add chemical substances with the right smell and taste, chemical structures with the right structure, and so forth. But, in doing so, the very notion of simulation would be defeated because, at the end, one would have a real apple and not a simulation of it.
In believing that games are becoming indistinguishable from reality, Musk is also operating under the notion that games are different from reality. If a simulated or virtual world was different from reality, it should be made of something the physical world does not contain. As mentioned above, it is a misunderstanding deeply rooted since the ancient Greeks. It is the belief that base reality is known through an immaterial level of appearances. Such a level is the immaterial level of our mental life of course. Descartes called it soul or thinking substance. Computer scientists call it software or models or… simulations. Of course, the word software refers to a very concrete collection of things that allow us to control objects and access information.
However, pace many philosophers, there is not a different level of reality that our computers access when they do number crunching. Computers have no mental life. Computers have no inner images. There are no simulated worlds inside computers. Inside computers, sadly, there are no simulated trees, spaceships, explosions, luscious naked bodies, or scary monsters. Inside computers, there are, boringly, just voltage levels that control physical objects called screens that, when viewed by human beings, produce the mistaken-yet-amusing belief that one is watching trees, spaceships, explosions, and the like. Yet—and this is key—to do so, a computer must control another object (commonly a screen) that exploits physical colors.
Consider virtual reality, for example. Next December a stampede of VR headsets will dominate our holiday shopping choices. And everybody assumes that VR worlds are only mental, that they are simulation. They are only virtual worlds; they do not exist in reality. Correct? Wrong. Inside those VR headsets there are small LCD screens that produce physical colors. There are actual moving pictures that change. The connected headphones produce actual sounds that are picked up by players' ears. And so forth.
The real Virtual Reality (pun intended) is made of physical (albeit tiny) things and physical phenomena. We misinterpret the tiny stereo pictures inside the screens of the VR headset as though they were, say, a giant DOOM imp, but they are not "immaterial": They are actual physical colors and pixels and LEDs. VR is not an immaterial world; it is rather a physical world we misinterpret. Like when we are appalled by a picture of a yummy Sachertorte that, while physical, is just an inedible surface of pixels. VR is akin to a super-technological magic trick in which beholders believe they see something they don't really see and that does not really exist.
The key point, of course, is the nature of the mind. In fact, when one dreams there is no computer screen on which colors are physically instantiated. So far, we are not aware of any brain screen on which mental colors are physically instantiated either. This is what Daniel Dennett called "The Cartesian Theater." But we have no idea whether dreams are like virtual reality simulation. Let us stress again the point: computer simulations are not immaterial. They rely and depend on actual physical properties: sounds from headphones and speakers, colors and lights from screens, movements, pulls from haptic devices. And so forth.
And yet, too often models, thoughts, simulations are taken to be immaterial things—or, to use Musk's words, to be above the base reality—as though we were still prisoner of the Cartesian division between a physical world (base reality) and a mental/immaterial world (the simulation).
Brains Are Not Computers
It is not by chance that the philosophical chickens come to roost when notions like mind and simulation collide. While it is no longer considered respectable to profess a belief in disembodied immaterial souls, many philosophers and scientists defend surprisingly similar views regarding minds, computations and simulations.
In this regard, Musk's statement was backed up by Nick Bostrom's influential 2003 paper, "Are we living in a computer simulation?" Bostrom summarizes his view in these terms: "A common assumption in the philosophy of mind is that of substrate-independence … Provided a system implements the right sort of computational structures and processes, it can be associated with conscious experiences. It is not an essential property of consciousness that it is implemented on carbon-based biological neural networks inside a cranium: silicon-based processors in a computer could in principle do the trick too."
We have no empirical evidence that computation, whatever it is, leads to conscious experience.
Bostrom restates, using modern philosophical jargon, the Platonic-Cartesian notion of a level of reality independent from the physical world—a reformulation of Putnam's thesis of multiple realizability. Bostrom says, "First, we formulate an assumption that we need to import from the philosophy of mind in order to get the argument started."
In other words, the simulated universe hypothesis requires one to adopt the computational stance. Very simply, this is the assumption in philosophy, cognitive science and AI that computations are sufficient for thought. Bostrom continues, "Suppose that these simulated people are conscious (as they would be if the simulations were sufficiently fine-grained and if a certain quite widely accepted position in the philosophy of mind is correct)." The "certain quite widely accepted position" is known as computationalism—which is the belief that consciousness is isomorphic with or caused by computations. It is nothing short of an article of faith, since we have no empirical evidence that computation, whatever it is, leads to conscious experience.
Bostrom asks to believe that consciousness stems out of computational processes. Yet, we know that many scientists are very skeptical about the analogy between brains and computers. For one, a recent paper by Robert Epstein on Aeon, "The Empty Brain," addresses this point and has stirred quite a reaction. After all, while brains can do computations, they are not computers. They are complex aggregates of cells that interact with the world by means of our bodies. We describe them as though they were computers, but they were not designed as computers. In fact, in many respects, brains would be awful computers and vice-versa.
Thus, Bostrom's thesis is suspicious and packed with unproven assumptions. We do not know whether a brain is a computer. We do not know whether a computation will yield consciousness. Actually, so far we have no evidence that computations in computers produce any consciousness. We do not know whether consciousness is really a level of reality distinct from base reality. This last point, if true, would, again, be an updated stripe of Cartesian Dualism. Why should we take for granted a thesis so problematic?
A Mind Needs The World
Finally, there is one more argument against the notion that we live inside a massive simulation of everything. Suppose that, against all odds, we were living inside a simulation made of something different from the now (in)famous "Musk base reality." If this were the case, the simulated world would be the only world we could access. Such a world would have the properties of the world everyone calls the physical world. Such a simulated world would therefore be identical with what everyone calls the physical world. The base reality would be utterly beyond our grasp and thus it would be, with an unavoidable conceptual twist, immaterial to us. It is a bit like that old joke: after centuries it has been found that William Shakespeare's plays have not been really written by William Shakespeare but by another man called William Shakespeare.
Either way, we live in a physical world, where physical is a catchphrase to refer to the world we live in. Once more, embracing an all-encompassing massive world simulation defeats its very nature. If the simulated apple replicates all properties of the apple, the simulated apple is the apple.
To recap, Elon Musk's argument—that a) once we had Pong, now we have Doom, therefore b) in the future there is a very good chance that we will live inside simulated worlds (and this might be already the case)—is unconvincing because nothing links b) with a). They are different things, both empirically and conceptually. The world we live in is made of real stuff. Simulations are things made of the same stuff. Musk's argument does not show that we are getting any closer to producing an alternative reality. Rather it shows that we are getting better and better at shaping the physical world.
In fact, games are becoming like little aquariums that flesh out with increasing accuracy a piece of the physical world. They are a bit like ultrasmart dynamic HD dioramas. In fact, dioramas are three-dimensional full-size or miniature models, sometimes enclosed in a glass showcase for a museum. Dioramas are physical simulations. A virtual world is like a diorama only that it uses electronic colored surfaces rather than wood or plastic scale models. A screen inside a VR headset is an amazing piece of reality that, like a superfast chameleon, reproduces all colors and shapes. It is not an immaterial figment of one's imagination. It's a piece of matter with colors, mass, and electricity interacting with your brain.
If a simulated waterfall is not wet, why should a simulated mind think or feel? A mind, unless one believes in disembodied souls, requires a brain, a body, and a world. A mind without a physical world is a myth. And a simulated world is a myth too. The fact is that all minds we know of, human minds and possibly animal minds, are embodied and situated: they have a body and they partake of the physical world. We have never met a disembodied mind. We always meet bodies in the world.
Riccardo Manzotti is a Professor in Psychology at the Institute of Human, Language and Environmental Sciences at the University of Milan, holds a PhD in robotics, and is the author of 50 papers on the basis of consciousness. His website is consciousness.it
Andrew Smart is a cognitive scientist and the author of two books, Autopilot: The Art and Science of Doing Nothing and Beyond Zero and One: Machines, Psychedelics and Consciousness (OR Books).