I used to love to play video games.
My favorite of all time was this one called The Secret of Mana, back in the old days of the Supernintendo. My brother and I never had a Supernintendo ourselves, mind, but we borrowed one once from a friend of his, and when we put in that magical cartridge…
The Secret of Mana was grandiose and mythical, like Zelda meets Willow. To my twelve year old brain, it was perfect. I used to feel like I was inside the game. Me and my buddies, the girl and the midget—whomping out bad-guy brains, gathering Mana, blasting from cannons, sneaking through tunnels, trudging through forests, castles, ghosthouses, seasons, Moon palaces and deserts, tundras, Witch lairs, Mushroom villages, and even flying over the world on our dragon friend Flammie. But, even among all of these marvelous textures, I never once thought that that little world might actually be real.
Wha-what? A video game—real? Well as far as ridiculous is concerned, it kinda is and it isn’t. After all, what do we think makes reality so much different from a video game? Is it just that reality is so much more complicated? Or is there something deep down that’s actually different about reality, some hidden level… like a spirit or soul layer. And if there is such an intangible, what’s to stop a videogame from having it, too? If we program feelings and quirks into Mario, will he actually feel them? The question becomes less screwy if I rephrase it: is there any way we could know? I mean, after all, I can’t even be sure that my best friend actually has feelings, as opposed to being an automaton. If Mario tells me he feels something and he’s convincing, how can I know if he’s lying?
If we didn’t have our own unique subjective views on the world, we wouldn’t even bother thinking about any of these questions. But we do have these views, so we must take seriously the possibility that other people, and maybe even simulations of other people—like in videogames, or more relevant to our purposes, in an attempt model all of the laws of physics in a computer (or in any other system that requires some abstract mapping in order to relate objects within it to objects in our world)—can also have subjective experiences. Such a simulation is of course only as good as the physics that go into it. But what if the physics that go into a simulation are perfect? What about the optimal case, in which a simulation represents the mechanics of our own reality so perfectly that the physical laws of the two worlds can’t be distinguished by any physical test: a test done here, in the real world, versus a test done there, in the simulated world. Well… then things become interesting.
If our question is whether all of the characteristics of reality, including subjective experience, could exist in a simulation—and that is our question—then the first obvious experiment within such a simulation is to produce a fully functioning human being, and then to test if it can experience the subjective. The simulated human either would or would not experience the subjective. If the human, whom we’ll name Turing, says that he doesn’t have any subjective reality—or if Turing simply displays no aptitude for any sort of regular functioning or thought—then we’ve concluded… nothing!
Actually, it is possible that Turing experiences all of the colors of the rainbow just as we do, but for some reason can’t express it to us; or that Turing experiences nothing, but describes the colors so accurately based on his operational programming that we become convinced that he does experience them subjectively. Well okay then… so much for the empirical question of being able to test whether a simulated person can experience the subjective. After all, we can’t even be certain that people in this world actually experience the subjective. All I can know is that I experience it. But what about the more fundamental ontological question of whether a perfect simulation of a person could theoretically experience the subjective, regardless of whether we could ever actually test it? This is still a question we can meaningfully ask.
So let’s take for a moment the possibility that, for some fundamental reason, no simulation of the laws of reality could allow subjective experience. What could be the cause of such a failure?
**********************************
(Reasons why even a perfect simulation might not be able to produce subjective experience)
1. Science is wrong (aka, Brain in a Vat): Is it possible that everything we know about the relation between mathematics and science and the functioning of nature is really a big hoax? That we misunderstand reality and are deluded, and that no set of rules actually govern the universe? In essence, this is the possibility that Occam’s razor—our most basic assumption, that the most elegant (or parsimonious) description of something is probably the correct one—is wrong.
Occam’s razor is the principle we use intuitively from our childhoods to understand the world around us. For example, because the sun rises every single day, we believe that the sun exists and that it takes on a regular pattern of rising and falling. This is a simpler explanation of the sun than, e.g., the suggestion that the sun rises every day only because some computer tells it to do so, and that one day (perhaps tomorrow), it will halt in midair. Because things around us tend to behave in a predictable manner, we believe that they are real, and we don’t create unnecessarily elaborate explanations for them. This is a basic premise of science as well, because science is merely the precise application of Occam’s razor to observed phenomena in the world.
If we scratch Occam’s razor (and hence science), anything is possible. The whole universe could be merely a movie run for your enjoyment, you being a brain sitting in a vat. Or all of existence could have been created on the spot two seconds ago, with all of our memories and experiences implanted. There are many inconsistencies and weaknesses to any such theory, because when we dispense with Occam’s razor, we enable a huge and improbable smorgasbord of implicit assumptions. For instance, if you are a brain in a vat, then what are the physical rules governing the vat? How did the brain get put in the vat? Where is the vat? A whole theology could be built based on answering such impossible questions.
And most importantly, no Brain in a Vat type theory actually helps us to answer any of our big questions: such theories merely redirect them. If we dispense with Occam’s razor, then instead of wondering how our universe can support subjective experience, we must ponder instead how, perhaps, …er… a universe composed of a brain in a vat, or something, can support subjective experience. And in this case, we’ll run into all of the same quandaries.
So let’s hold on to Occam’s razor, and move forward to more plausible, or at least more useful, scenarios.
2. Humans have souls, and souls can’t be simulated: Is it possible that humans have souls, and that souls are what experience subjectivity? Descartes formalized the soul concept with his theory of substance duality, which claims that there are two fundamental types of stuff in the universe: normal physical stuff, and soul stuff. The soul stuff affects the physical stuff, but does not obey its rules. Soul stuff is, in essence, a window into the world without Occam’s razor. Unfortunately, this idea has many problems. For instance, if the soul can interact with the body, what prevents us from measuring its effects with our instruments and then incorporating those rules into physics? We would simply expand physics to include soul stuff. Or perhaps we can’t do this, because reality is fundamentally as disordered as the soul layer. If this is the case, then physics has no fundamental meaning, and we are back to the ‘brain in a vat’ scenario.
Even if the soul theory is true, we still wouldn’t know whether a computer simulation would be able to produce subjective experience. All we would know for sure is that souls, the source of subjectivity, can couple with human brains. Whether a simulation can support subjectivity would depend on whether souls can also couple with simulated brains. Soul theory as it stands can say nothing on this matter, so the determination of whether simulations have souls would have to be added as an axiom to the soul theory, rather than being a derivative of it. We would be better off, therefore, just going ahead and guessing blind about whether simulations can experience subjectivity, since the soul theory itself sheds no light on the issue.
3. Subjective experience is excluded by Godel’s incompleteness theorem: One of the most remarkable logical proofs ever written is Godel’s theorem, which casts doubt on whether a full simulation of reality could be done even in theory. Godel’s theorem proves that for any logical system at least as complicated as natural numbers (aka, positive integers), there exists no logical framework to describe the system that is both complete and consistent: complete meaning that we can derive any true statement from basic principles, and consistent meaning that it is impossible to derive two statements of any sort that will conflict with each other. Any logical system fully describing natural numbers will include arithmetic (e.g., 1+1=2), as well as even more basic and seemingly obvious properties such that for any number n there exists a ‘next’ number, n+1.
Although natural numbers are the most intuitive type of math, it turns out strangely that fully formalizing the way they relate to each other is provably impossible. There must be holes in any consistent description, and any complete system will lead to jarring inaccuracies. What this means practically, for us, is that fully modeling the laws of the universe is literally impossible—unless, that is, the universe itself holds inconsistencies in the way laws of nature interact with each other, which would lead us back again to the world without Occam’s razor, or unless we believe that somehow the universe is categorically simpler than natural numbers (although how could that be?!). The question then becomes: even if we can’t fully model the universe, are the properties of the universe that we can model sufficient to simulate subjective reality? There are good reasons to think that they aren’t sufficient, because of the similarly self-referential nature of both Godel’s proof and consciousness. But while we can guess, we really can’t say anything for sure.
4. The stratum of the universe matters: Simply put, this is the idea that there is something about the reality we are stitched out of that is fundamental and cannot be modeled. This isn’t to say it can’t be modeled mathematically, but rather that even a perfect mathematical description would fail to capture certain properties because it matters what kind of stratum the model is implemented in. Many scientists would argue against this theory, since math seems at least theoretically able to capture anything that nature throws at us (sort of… see the last point, about Godel’s theorem). However, the one area that math and science have said nothing about is the actual experience of subjectivity—which, of course, is our topic. So how might subjectivity arise in the stratum theory?
The stratum theory states that the fabric of nature itself provides a sort of mapping between what can be measured objectively, and what can be experienced subjectively by the same piece of matter (keeping in mind, of course, that for a full subjective mapping to occur, the matter in question must also be arranged in a suitable way, such as how it’s arranged in a human brain). Because a simulation won’t be implemented in the right stratum—namely, in the stratum of our universe, with its quarks, atoms, molecules, forces, and dimensions (as opposed to, for instance, the stratum of a computer processor)—it may not yield the proper mapping, and thus might not be able to create subjectivity. This is to say, reality is like a movie projector, where the equations we can use to describe nature are like a filmstrip, and the subjective experience is like the projected film. (Note, this analogy is not about the experience of viewing the film, but rather about the projection of the film itself: as far as we are concerned, there could be nobody in the theater at all, but the movie would still be properly projected, and hence subjectivity would be created.) If we perfectly copy a film into some other format, like onto a flash drive, this would be like perfectly mathematically modeling reality. Even doing so, though, we would not have created the full subjective reality (the projected film) in a simulation (the flashdrive). To do that requires the projector and an actual filmstrip—the proper stratum. The stratum by its very nature provides subjectivity, whereas implementing reality in a simulation would be like trying to view a movie stored on a flash drive by shining light at it (which obviously would not work).
Why is the stratum theory more satisfying than the soul idea? It’s because it’s much simpler. First of all, we have strong evidence that some sort of mapping exists between subjective and objective experience, as it appears from all scientific inquiries that the physical brain and the subjective mind are one and the same, and yet we know they are experienced entirely differently. The brain is projected, so to speak, into ‘mind space,’ just as a filmstrip is projected onto a movie-screen. The stratum theory is therefore an explanation of our definite experience that uses as few and as simple assumptions as possible. The soul theory, on the other hand, assumes a serendipitous connection between an unbounded soul reality, which can have any number of unexplainable properties, and the ordered reality we see about us. It permits many more assumptions than the stratum theory. The soul layer is considered active, whereas in the stratum theory, the projected image is primarily passive.
One part of the stratum theory—the existence of a mapping between our subjective experience and some part of our objective body—feels self evident, because our objective experience of our brain (for example, seeing it during surgery, or via MRI) is radically differently than our subjective experience of being our brain. Our question about subjectivity, then, may be reframed: how does the mapping between the subjective and the objective arise, and does it require a particular stratum? For example, would mathematical equations that perfectly simulate reality be enough to produce subjectivity? Would these equations even need to be implemented? If they don’t, then will we create a true reality, with subjective experience and consciousness, if we write all the equations down on paper and solve them by hand? What if we don’t even write them down, but simply know they exist? Or what if we take this idea to its most logical extreme and just prove that such equations could exist, even if we will never actually know them?
If these last few ideas seem preposterous, then you have tacitly assumed that the stratum does matter. If simulating equations on paper or in a computer, rather than them simply existing, is required for them to be actualized, then really what we’re saying is that we need to borrow some part of our universe’s stratum—such as its physical dimensions or even the flow of time—to make a simulation that has its own subjective reality. We must then ask not if we need a stratum, but rather whether subjectivity requires the parts of our universe’s stratum that cannot be passed to a simulation.
**********************************
So those are a few reasons (and there are others) why it might be fundamentally impossible to simulate the functioning of reality perfectly, or at least well enough to produce subjective experience. But if it turns out that we can simulate the functioning of reality perfectly… what does it then mean for our understanding of nature?
First of all, if we find we can simulate reality perfectly and with no fundamental losses in its character, we will gain very strong evidence for the possibility that we also are living in a simulation. After all, once we’ve exhibited the possibility of alternate but equivalent layers of reality, why would we expect that the one that we live in is the most basic one? There could be a layer simulating us, and another beneath that one, and on and on to infinity. But even this layering supposes some sort of stratum, because the layers are dependent on each other insofar as they are implemented within one another.
There is another even simpler explanation of how our multiverse is structured, again assuming that we can simulate reality perfectly. You can decide for yourself if you find this explanation intellectually satisfying. This explanation remarkably requires no specialized stratum at all. It is Max Tegmark’s Mathematical Universes Hypothesis (MUH), which states that every mathematical structure is real. In MUH, a triangle is a viable universe, as is a rhomboid, or the natural number system, or even, in some way, The Secret of Mana. Of course, merely existing does not mean that a universe can support subjective experience. First of all, MUH presupposes that subjective experience can derive directly from mathematics—a tricky proposition, as we’ve seen—and secondly, a given universe within MUH might or might not contain subjectivity depending on whether it is sufficiently complex. Of course, we know for sure that our own universe is in the class of mathematical constructs that allows subjectivity. This leads to the remarkable and strange fact that Tegmark’s hypothesis can actually be tested… sort of… using statistics. If indeed any mathematical structure “exists,” then there probably also exist infinite mathematical constructs above the complexity threshold that enables self awareness. We would therefore expect that our universe is among the more “probable” mathematical structures that can produce consciousness. At least in theory, this could be tested, if we gain sufficient understanding of physics to understand what mathematical principles are required to produce subjectivity.
So there you have it. If we look over all of our theories and crystallize them, I believe that we come down to one fundamental divide: do we take a stratum interpretation, in which the substance that our universe is implemented in has a special and unique nature, or do we take a non-stratum interpretation, which naturally leads towards the simulations-within-simulations or the MUH theory. Either one has different consequences on whether we can create simulated humans who can experience their own subjective reality. The most interesting part is that with the current state and the pace of technology, we might be actually exploring these questions in real simulations pretty soon. Maybe we’ll even be able to breathe real life into The Secret of Mana. And if we can do that…
P.S. A blog that I follow, which has many insightful and sharp posts about many of the points I’ve highlighted, is: http://www.reenigne.org/blog/category/philosophy/. Enjoy!
No comments:
Post a Comment