Is it possible to prove that the universe either is or isn't a simulation?



Can it be philosophically proven that the Universe either is or is not a simulation? If someone was in a simulation, could they tell? What would the differences be between a simulated universe and a "real" universe?


Posted 2015-05-29T21:50:22.220

Reputation: 161

Question was closed 2015-06-01T20:56:35.987

Thats certainly relevant, but my question is less about whether we exist, and more about whether we exist in something real versus something simulated, or even if there is such a concept of "real" vs "simulated" – DCIndieDev – 2015-05-29T21:53:16.540

Although it's not the same question, I think you may find an answer there. In any case, also relevant: How does one know one is not dreaming?

– None – 2015-05-29T21:56:16.730

3We cannot be dogmatic in philosophy, as we sometimes are in science, where phenomena or explanatory models are guiding presences, and concepts can be defined quite loosely. It may be surprising to you, but we cannot understand what you mean by "universe", "simulation", and "real", without any - lenghty and well-reasoned - characterization. – André Souza Lemos – 2015-05-29T22:36:50.027

Well, it's impossible to disprove it, certainly. However, it isn't relevant. If we did or did not, we would still be restricted by the same rules that we are now. – Goodies – 2015-05-29T22:52:54.760



Although far from agreed upon by all philosophers, DesCartes' cogito - the famous "I think therefore I am" - was a result of him attempting to prove that the universe was not a simulation.

He starts out by postulating that everything he sees, hears, feels etc, might not be real, but was actually an illusion created by an evil demon, trying to make humans believe they were part of a real world when they weren't. They didn't have computers or Keanu Reeves in the 17th century, so illusions created by demons has to do in lieu of simulation.

The demon could make him doubt that anything he sees is real, or that anything he tastes is real, etc...

However, while he can doubt everything else, he can't doubt the fact that he is actually doubting anything. Because doubting that he has the ability to doubt is in itself doubting.

So he has established that at least the act of doubting is real. But for there to be doubting there has to be thinking. So thinking is real as well. But there has to be something to do the thinking. So now he has established that the soul (the "I") is real. Hence "I think, therefore I am".

DesCartes claims that this proves with certainty that the soul is real and not a simulation (or demonic illusion).

Hope this helps.

Alexander S King

Posted 2015-05-29T21:50:22.220

Reputation: 25 810

As a simple proof that in this interpretation the good René was wrong, a simulation of him would have come to the same conclusion. However I'm not sure that it's likely to be a correct interpretation. On the third hand, the middle ages' thinking about angels as infinitesimal sized thinking machines (with an infinity of them fitting on the head of a pin) was extremely sophisticated, so nothing can really be dismissed. – Cheers and hth. - Alf – 2015-05-30T10:13:49.560

while I do not believe that a Turing equivalent computation can reproduce a mind, this does not strictly exclude the existence in principle of some other kind of super "computation" which can produce a mind, and that we are all, or at least the mind reading this comment, part of that simulation by some unknown alien race... – nir – 2015-05-30T11:52:59.220

1@nir: Regarding "while I do not believe that a Turing equivalent computation can reproduce a mind", any finite physical system can be simulated to any desired degree of accuracy by a sufficiently powerful computer. Thus a bear's brain can in principle be simulated, with recorded inputs applied, in sufficient detail to not create any deviation of responses over a stretch of, say, two months. One may argue that the mind presence there is only relative to the inputs, but, disregarding the ethics of it, use real world inputs. Thus, the idea of no mind here is belief in something supernatural. – Cheers and hth. - Alf – 2015-05-30T12:37:12.063

@Cheersandhth.-Alf, the claim about a finite physical system sounds like Deutsch's principle, which I am not aware is generally accepted - see the SEP article on the Church Turing Thesis for a discussion of misinterpretations of the original CT thesis; while it makes sense that turing equivalent computation may approximate the evolution of the physical laws that we (!) formulate, you should consider that these laws are themselves just approximations of nature (see Feynman).

– nir – 2015-05-30T13:56:00.923

@Cheersandhth.-Alf, as one example, out of my sleeve, while you can use physics and computers to simulate the formation and evolution of a galaxy, can you similarly compute what is going on inside a black hole? at the singularity point? so, the bottom line is that we do not know if a bear' brain or a humans' brain, can be simulated by a turing equivalent computation. my bet is it cannot - and consciousness is in fact in my opinion the evidence. – nir – 2015-05-30T13:58:33.983

The existence of something that can not be simulated, because we don't know anything about its innards, does not prevent simulation of something we can know everything about at the level of interest for the simulation (namely, the individual molecules). So that's one fallacy. Secondly, the idea that a simulation of a brain, reproducing the ordinary behavior, does not reproduce a mind in there, is of course a matter of belief. But I would like to point out that it's a racist belief, that can engender racist behavior. That's what often happens when religious ideas are mixed up with things. – Cheers and hth. - Alf – 2015-05-30T14:54:44.460

Regarding Deusch's principle, I have yet to see any explanation that doesn't indicate that it's not just a definition of the term "finitistic" (or whatever it was). I fail to see how lack of acceptance of a definition of a word or term, affects anything. Taking the comment at face value you might equally valid argue that since my comments are written in English they are similar to something else you've seen also written in English. It is absurd, yes? But I think maybe what it says is that you misunderstood my use of the word "finite", which was in the sense of "finite mass and extent". – Cheers and hth. - Alf – 2015-05-30T15:00:43.690

OK, maybe that was not clear. So: when I emphasized any desired degree of accuracy it was to make clear that one does not need the idea of an at bottom discrete (digital, symbolic, finitistic, whatever) reality. Reality can be infinitely detailed and yet it can be simulated to any desired degree of accuracy. Then, if a computer model behaves exactly like you, over several months, I would tend to treat it like you, except where the properties differ (e.g., turning off and on again is not as drastic as with the biological you, because the simulation can be resumed). I would say it has a mind – Cheers and hth. - Alf – 2015-05-30T15:22:53.450

@Cheersandhth.-Alf, to explain the black hole analogy, from my point of view, consciousness is beyond Turing equivalent computation as the inside of black holes is beyond measurement; as for a simulation of a brain, since I believe brains create consciousness, and consciousness is beyond Turing equivalent computation, my conclusion is that brains are beyond Turing equivalent computations; no Turing equivalent physical simulation of a brain will produce a thinking person; in that analogy, simulating a molecule or 1000 molecules is like simulating a planet orbiting a sun; – nir – 2015-05-30T21:45:38.817

@Cheersandhth.-Alf, can simulate the entire galaxy, but that does not mean that you can comprehend or compute the singularity at it's center; in the case of a galaxy, we do not really care, since it is basically stars and gas orbiting a center, one way or the other, but in the case of a brain, it will never get off the couch; people are talking about an inevitable "metaphorical" singularity of AI; I say open your eyes, a real singularity has happened a hundred thousand years ago in the human brain - we call it a mind. – nir – 2015-05-30T21:46:33.263

@Cheersandhth.-Alf, as for Deutsch, I wasn't complaining about the word "finite" in his claim, but that people basically take it to mean that reality is computable, and they believe such a claim has the same status as Turings original thesis which had nothing to do with it, but was actually about Turing machines being able to simulate the job done by human "computers" - i.e. clerks doing mathematical computations with pen and paper. – nir – 2015-05-30T21:51:48.180

@nir: Well, I think can't do more about the belief that minds are supernatural. Except that I ask you, should you live for 30 years or more, to treat machine intelligences with the same respect you would have afforded entities with minds. Never mind whether they "actually" have "minds" (here I'm waving my hands frenetically). Just consider them a kind of possibly handicapped creatures who, in a mindless way of course, are really smart, and well able to understand and react to others' treatment of them. If the notion of mind you have is restricted to humans, also extend this to animals. Thanks. – Cheers and hth. - Alf – 2015-05-30T22:10:53.587

@nir the Church-Turing thesis applies to any finite discrete process. The only way we can achieve hyper-Turing computation is if we somehow can (a) perform an infinite amount of a steps in a finite time, (b) or if we had access to continuous variables (i.e. could measure things with infinite precision). (a) is intuitively impossible. What Deutsch noticed is that per quantum mechanics and the uncertainty principle, (b) is impossible as well. It is thus very straightforward that any hyper-Turing computation is impossible. I am puzzled by why people consider him to have misinterpreted Turing. – Alexander S King – 2015-05-31T03:46:09.533

@nir That "Misunderstandings of the Thesis" in the SEP article is an embarrassing load of bull, whoever wrote it has no business writing SEP entries (and in fact has me doubting the quality of everything on SEP now :-( ). Turing's thesis is a direct response to Hilbert's Entscheidungsproblem. As such it is about any finite step procedure. This statement is [Turing did not show that his machines can solve any problem that can be solved "by instructions, explicitly stated rules, or procedures"] is flat out wrong. – Alexander S King – 2015-05-31T03:58:02.250

@AlexanderSKing, the author of the SEP entry on the Church Turing Thesis is the philosopher Jack Copeland, and according to the Wikipedia it seems that Alan Turing is his forte. If you have good criticism of the entry I would really love to read it, but labeling it as BS or flat out wrong without explanation does not count. how about in chat or as a separate question?

– nir – 2015-05-31T06:35:43.553

@Cheersandhth.-Alf, I don't believe the mind is super natural but that physics and Turing equivalent computation are sub natural. you can think of it as cave men's fire; I'm saying you can't light fire with Turing computation, not that it is impossible to find a way to artificially light fire. (please note that this metaphor is not at all about the wetness vs simulated wetness argument!) – nir – 2015-05-31T09:18:50.613

@AlexanderSKing: Just as one can reason about the math here without reference to an authority (yes, JC is trivially wrong here), so one can reason about whether a device that only performs input at startup, and output at termination, can model an ongoing process. No, it can't. Processes with input can be modeled as functions over infinite sequences (I got that from dataflow languages in functional programming), but that's not a Turing machine. And that knocks the feet from under all the TM-based arguments. – Cheers and hth. - Alf – 2015-05-31T10:22:43.737

@Cheersandhth.-Alf, I don't immediately see a problem with having TM simulate an ongoing process; here are two options out of my sleeve: a) simulate a self contained environment, e.g. a room with two guys talking to each other; b) have the TM compute the simulation one step at a time, where the evolving state of the system is carried from the output of one computation to the input of the next, together with any "interactive" input you wish to inject into the system; I don't see how Copland is trivially wrong, and I would love to read a satisfactory and respectful criticism. – nir – 2015-05-31T14:15:39.290

@nir: First, I note your request for a "respectful" critcism and the implied slight, as well as its purpose. A TM is a simple stateless function (input → output). Letting it simulate two minds before producing its output is not problematic, it's just irrelevant to TM-based arguments. The idea of connecting up a series of some trillions of TM's is relevant, because it's one way to do the job of representing an evolving intelligence. The system's Gödel statement then changes at each state update, i.e. for each machine succession in your model, including after a question has been posed. – Cheers and hth. - Alf – 2015-05-31T15:01:51.960

@Cheersandhth.-Alf, no slight was intended; my point was that claiming Copland is "trivially wrong" or that his essay is "load of bull" is not a convincing argument; can you explain why using a TM to simulate a self contained environment is irrelevant to "TM-based arguments"? I did not suggest chaining trillions of TM but running the same one again and again on the same tape; note that this "mode" is arguably analogous with the self contained simulation "mode". – nir – 2015-05-31T20:51:01.973


@nir I had already explained in my previous comment: The Church Turing thesis applies to ANY finite discrete algorithmic process. Keywords: finite and discrete. Don't take my word for it. In fact the author of another SEP article on Turing, Andrew Hodges disagrees with Copeland: and Church's student Martin Davis disagrees with him as well And here's MIT professor Aaronson:

– Alexander S King – 2015-05-31T21:43:02.753

@Cheersandhth.-Alf "Processes with input can be modeled as functions over infinite sequences (I got that from dataflow languages in functional programming), " Sounds interesting. Can you elaborate? Note that if you are modeling it on an a digital computer, then it is running on a Turing machine, since there is equivalence between digital computers (Von Neumann machines) and Turing machines. – Alexander S King – 2015-05-31T21:47:21.407

@AlexanderSKing: One (the first?) language was called Lucid: “Each variable is an infinite stream of values and every function is a filter or a transformer”. And yes, they're equivalent in processing ability: any function (number → number) that can be computed this way can be computed by a Turing machine and vice versa. "nir": The equivalence in what can be computed does not extend to an equivalence in possible behaviors with respect to i/o, because, simply, Turing machines don't have i/o, they're intentionally handicapped.

– Cheers and hth. - Alf – 2015-05-31T23:11:31.773

@nir: We've gotten far off original track now, but following up anyway: the idea of using the same TM repeatedly is not unworkable, it would just effectively have to be a universal TM. We could model it as a function ((input, state) → (output, new state)). The starting state can be a number K, and the new state a number L. Then this application of a universal TM is equivalent to the action of two ordinary TMs that can be inferred from K, that compute resp. the output number and the new state, each with Gödel statement depending on K. I suspect this was the original fallacy in the 1940's. – Cheers and hth. - Alf – 2015-05-31T23:33:16.650

1@AlexanderSKing, thanks for the links! I'd say they were worth the heresy of 19 back and forth comments... – nir – 2015-06-01T00:54:00.770

@Cheersandhth.-Alf, and Alexander S King, According to Andrew Hodges in the SEP article on Alan Turing that in his last years Turing was concerned with the consequences of quantum mechanics on the computability of the brain: "Probably Turing was aiming at the opposite idea, of finding a theory of the reduction process that would be predictive and computable, and so plug the gap in his hypothesis that the action of the brain is computable."

– nir – 2015-06-01T18:35:03.307

@nir: Thank!. I think this speculation about Turing throws some more doubt on the reliability of SEP. On the one hand it's well worth noting that just about every genius on record has publicly stated some nonsense ideas, and Turing is infamous for embracing ESP (extrasensory perception), and less known, a racially segregated society. But on the other, with Turing it was all rational: his attitude to ESP very reluctant, but forced by apparent evidence. With Penrose (cited as authority in the piece), it was instead very irrational. You might find it amusing to read Moravec about that. :) – Cheers and hth. - Alf – 2015-06-01T22:33:25.467


– Cheers and hth. - Alf – 2015-06-01T23:16:34.313

@Cheersandhth.-Alf, I don't find it amuzing. The SEP article states "widespread disagreement" with that particular penrose's belief; I submitted an observation above, and in response you write the SEP is unreliable, Penrose and Turing believe in nonsense and Turing is a racist... you also accused me of holding to racist ideas above (why? because I'm not a reductionist?) ; I think it can be characterized as ad hominem; to say the least, I don't believe this is a useful method of discussing things in a philosophy website. – nir – 2015-06-02T09:36:03.953

@nir: Just some corrections. (1) I have not written that the SEP is unreliable, but it's a fact that it's incorrect in one area, that of JC's discussion of Turing's proofs. So the quality assurance is proved a bit lacking, but I don't know how bad/widespread it is. (2) I haven't written that Turing was a racist. But he evidently held to a belief in a racially segregated society, talking about “friendliness …between white man and white man, or between black man and black man”. (3) You're wrong that facts are ad hominem attacks. They're not even in that direction. They're the very opposite. – Cheers and hth. - Alf – 2015-06-02T14:06:50.893


You can make a good start on proving that reality is simulation, by doing certain physics and possibly astronomy (1)experiments that would force the simulator to expend ridiculously high resources to maintain consistency. But then the simulator, if it has (2)snooping ability, can just restart from an earlier time with some slight variation, where these experiments would hopefully (for him/her/it) not occur. I.e., you can't even in principle successfully battle simulator's censorship.

Still, (3)science fiction stories have been written about people doing exactly that, battling the simulator more or less successfully, such as Vernor Vinge's “The cookie Monster”. Informally, I find that the presentation of the story as pure text with arbitrary emphasis added, makes it hard to read. One does not perceive the structure of the piece without actually reading all of it: it's not easily accessible! Let it be more formally stated, though, that Vinge's references to Usenet communications and a galaxy filled with Norwegian names, in some of his books, makes him an author to be trusted. For it is obviously true that the galaxy will end up with lots of Norwegians and an old Usenet-like communications network, and still Vernor Vinge's the only author who has described it that way.

There is another way of viewing this, however. One can take the idea of cosmic censorship and say that it indicates a simulation, i.e. that it's likely a manifestation of just simulator's censorship. But as I see it that's purely nonsense associative reasoning, for cosmic censorship is about logical consistency, not about the computational cost of consistency.

Yes, I know, this answer's a collection of rambling thoughts & associations.

But IMHO that's the best you can get: attempts at reasoning more clearly about this can, I think, be dismissed out of hand, for it that was possible, then it would have been well known.

1) I don't recall exactly those experiments, sorry; I only remember reading about it, but no doubt they involve exponentially increasing effects (i.e. chaos, in its mathematical meaning, more informally the butterfly effect) and inspecting ever more fine details.
2) I almost used the acronym SDS, Heinlein's Super Duper Snooper, with a footnote explaining it, but then I thought better of it.
3) Can't leave this without mentioning Charles Stross' “Accelerando” universe and “Glasshouse” novel.

Cheers and hth. - Alf

Posted 2015-05-29T21:50:22.220

Reputation: 587


As with most things in philosophy there are no universally agreed upon proofs. There are arguments both for and against the skeptical hypothesis that reality could be a simulation.

As some people have mentioned Descartes malicious demon is a good place to start. But as far as proving the possibility or impossibility of a simulated reality I would recommend reading Hilary Putnam's "Brains in a Vat" In which he puts forth an argument against the skeptic hypothesis based on his semantic externalism.

In regards to the differences between a simulated universe and a "real" universe, it gets a lot more complicated. There are many fields of philosophical enquiry (i.e. ontology, metaphysics, epistemology) and a variety of theories within each, which would determine if there where any differences and if so what they would be.

An example, without getting too technical.

If the question is what differences would there be between a "real" universe (as is generally accepted) and a simulated universe (such as the brains in a vat scenario), then depending on the areas of concern certain differences or similarities would arise.

For instance it could be said that our experience of reality is dependant on various inputs. In the "real" universe these would be sensory inputs stimulated by external objects and their properties. In the simulated universe these inputs would be generated by the simulation.

So In the brain in the vat scenario compared to our actual reality. These inputs, which are generated by a simulation, would be fed directly into our brain. Whereas in our "real" universe these inputs generated by "real" objects and properties, would enter our brain via sensory organs.

In this scenario the differences would be:

  • the source of the inputs
  • the mode of entry

Furthermore it would follow (I would argue, fairly uncontroversially) that our experience in both cases would be no different. (Of course this depends on the quality and content of the simulated universe.)

I hope this helps to illustrate one such method of investigation, and how it is dependant on various premises (i.e. How we experience reality). If we were to reject these and put forth alternate premises our investigation might yield different results.


Posted 2015-05-29T21:50:22.220

Reputation: 21

Re "would", is that "would like to" or "can"? If the latter, are the results known? If the former, desirable but not done, what does that mean? – Cheers and hth. - Alf – 2015-05-31T12:52:26.630

"Would" as in "would follow from a particular framework". I guess I meant a combination of both of your interpretations of "would". Meaning that a particular theory "would like to" determine if and what differences there would be, and it "can" do so within a particular framework. I have edited my answer by including an example, which I hope will help to illustrate my point. In regards to your last sentence ("if the former...") I am not sure what you are asking. – Method – 2015-05-31T23:02:01.163


Nothing occurring in the physical world can be proven absolutely because reality itself must be assumed to be true. Therefore, it is possible that the universe is an imitation.

However assuming reality is fact does not mean that we cannot observe natural patterns to the point where we can predict certain outcomes with 99.999% probability.

The probability is what we have to work with. Certainly, the trend is your friend until it ends.

Ronnie Royston

Posted 2015-05-29T21:50:22.220

Reputation: 667


when some immaterialist claims that reality is no more than a simulation, you could react something like Samuel Johnson and stomp on his/her foot and say "I refute you thus!".

that should settle the issue.

robert bristow-johnson

Posted 2015-05-29T21:50:22.220

Reputation: 516