I was just listening to an interview with David Chalmers where he opined that if one could accurately simulate the brain, consciousness would arise in the simulation. Are there any other instances where that happens or anyone expects it to happen? If you simulate the motion of the planets, it doesn't give rise to gravity. If you simulate the random motions of molecules in a gas, it doesn't give rise to heat. If you simulate the action of a lever, it doesn't create force. Are there any other examples where a process X gives rise to a physical effect Y, and simulating process X on a computer also gives rise to an observable effect that is not simulated? What I mean is an effect that can be observed by someone who does not observing the output of the simulation. Note that consciousness can be observed by the consciousness itself without observing the output of the simulation, if any.

  • A simulation is essentially an abstract model of a system. Abstracting implies losing features (e.g. a house plan is an abstraction of a house), and models are always incomplete (e.g. a model has never the same physical attributes as the original). If you would not simulate, but replicate the system, there's no reason for the replica to perform in a different way than the original. Physical laws are universal.
    – RodolfoAP
    Commented Jul 1, 2023 at 7:55
  • @RodolfoAP, if you were to actually build an artificial organic human brain, it is reasonable to assume it would be conscious. Chalmers was talking about simulating one using electronics. Commented Jul 1, 2023 at 10:41
  • Cosmological simulations have become n essential tool to test the consequences of variations in parameters, & to explain large scale features of the universe like the Cosmic Web. Turbulence, in fluid simulations. Feedback, & all kinds of emergence, can be demonstrated in simulations.
    – CriglCragl
    Commented Jul 1, 2023 at 10:41
  • 1
    If you simulate gravity, your computer does not attract nearby bowling balls (except to the extent that its mass did before you ran the simulation). Chalmers is making a claim without evidence.
    – user4894
    Commented Jul 1, 2023 at 22:25
  • @DavidGudeman "Chalmers was talking about simulating one using electronics": obviously. That's the sense of my comment, any simulation is incomplete (ergo, the result is flawed in multiple dimensions). It can only be complete if the physics are identical.
    – RodolfoAP
    Commented Jul 2, 2023 at 4:31

10 Answers 10


I was just listening to an interview with David Chalmers where he opined that if one could accurately simulate the brain, consciousness would arise in the simulation. Are there any other instances where that happens or anyone expects it to happen?

Yes: computations.

Imagine that I have a handheld calculator that's capable of adding numbers together. I can enter "3 + 6 =" into the calculator, and it will show me the sum, which is 9. Suppose I also have a desktop computer with a program on it that simulates the handheld calculator. If I type "4 + 7 =" into the computer program, then it will simulate what the calculator would do if I entered the same expression. Specifically, it will simulate the process by which the calculator adds 4 to 7. But in the course of simulating the process of adding 4 to 7, my computer will actually add 4 to 7.

It may perform the addition the usual way (using an "add" instruction that its CPU executes), or it may do it in an unusual way (such as simulating each of the individual logic gates that form the calculator's adder), but no matter which method it uses, the computer is, in fact, adding 4 to 7.

Likewise, suppose that I have a computer game which is written for x86-based computers, but I have an ARM-based computer, but my computer has a program which simulates an x86-based computer. If I load the game into the simulator, then my computer will simulate the process of running the game, and in doing so, my computer will actually run the game. It will run it in an unusual way (because the usual way is to run the x86 instructions directly on an x86 CPU), but it will run it.

If consciousness is a type of computation, then the same argument goes for consciousness. To be explicit, the argument would say that if a computer simulates performing the kinds of computation that comprise consciousness, then the computer will be actually performing the kinds of computation that comprise consciousness.

(If, on the other hand, consciousness is not a type of computation, then it seems like consciousness definitely cannot arise on a computer, because essentially the only thing that a computer can do is compute.)

  • Also: proofs. A proof in a simulation is still a proof (probably).
    – Pablo H
    Commented Jul 3, 2023 at 15:58
  • As a trivial example, many people build logic circuits in minecraft, dwarf fortress or other building games. These are simulations of a computers logic gates, which produce the same computation as real world ones.
    – lupe
    Commented Jul 4, 2023 at 12:50

Relationships between properties are "real things" that exist in simulations. For instance, a simulated banana does not have a real surface area, but it does have a real ratio of simulated surface area to simulated volume.

If consciousness maps to a kind of relationship e.g. between senses, physical structures, and electro-chemical patterns, then a simulation of senses, structures, and patterns can have real consciousness. If consciousness maps to a process or to a set of physical system properties, then a simulation can only have simulated consciousness.

Various thoughts:

For all I know, simulated human consciousness might be an adequate antecedent for real computer consciousness, since the computer itself is obviously a real thing, and we don't really know what consciousness is or where it comes from, just that humans seem to have it. However, the claim is that the simulation would have consciousness, not that the computer would have consciousness because of the simulation, so that's not really relevant.

Accurately simulating the brain, if what is meant is completely simulating the brain, is fundamentally unlike any of the kinds of simulations you're comparing it with - and unlike any kind of simulation ever done. At the moment we can't completely simulate an arbitrarily small region of perfect vacuum, much less any traditional objects. Plus, no simulation of the brain could be complete without a simulation of its inputs from the rest of the nervous system, which can't be completely accurate without an equally complete simulation of its own chemical inputs, and so on. Perhaps what is meant is "simulate the brain well enough to get consciousness as an emergent phenomenon in the simulation", but that appears tautotological.

Although many very smart people think not only objects but the whole universe can be simulated, I suspect it's in-principle-impossible to completely simulate anything, short of finding or building (or procreating) a real physical instantiation of the process you wanted to "simulate" and really doing the experiment.


I think you're misrepresenting the argument a bit. I'm a bit puzzled as to what exactly you mean by simulation here. You wouldn't simulate the action of a lever to create a force... you'd simulate forces to simulate the action of a virtual lever. We want a bottom-up simulation. Why not have a simulation that exactly simulates the elementary laws of physics? If you hold to physical reductionism, then imo the natural deduction is that all simulations are equivalent to their non-simulated counterparts.

What do I mean by physical reductionism. 1)All matter in the real world obeys the laws of physics. 2)The only facts are those given by physics... mass, charge, field strength etc... These are simply variables in relation to each other via equations. It's mathematics, nothing else, and it can be simulated on a computer.

If physical reductionism is true and you simulate the real world in a computer... then the simulated humans will digest their simulated food. All the behavior will be the same.

So there's an isomorphism between the simulated world and real world if you hold to physical reductionism

So if you hold to physical reductionism... consciousness is completely derived from basic physical laws. ie: there's nothing more to it than the elementary particles obeying the laws of physics. We can simulate this in a computer. If a human being's consciousness IS simply the behavior of elementary particles... then simulated human's consciousness IS simply the behavior of simulated elementary particles.

If there is something different between the two situations... ie: the real world has something we haven't included or cannot include in the simulated world... then the only way around it is that physical reductionism is false.ie: there's something in the real world other than the entities of physics. To me this is obviously the correct solution, and as a result leads to a dualistic worldview. This is the most common sense way of viewing things.

Searle states that a simulation isn't the same as the real thing to make the argument that computers aren't conscious, but he doesn't explain why exactly. He seems to insist on physical reductionism, in which case it's hard to see what a simulation is lacking.

By physical reductionism, the real world is just a big computer.

. If you simulate the random motions of molecules in a gas, it doesn't give rise to heat.

What do you mean by this exactly? Is heat the random motion of molecules? We have a certain mathematics being implemented in the real world... and if it's also implemented in the computer... what exactly is the difference? eg: why does the relationship between a human and heat differ from the relationship between a simulated-human and simulated-heat?

The problem is the people insisting on physical reductionism is true and at the same time insisting that simulations can't be conscious. That doesn't make sense. It's contradictory.


This generally happens with informational resources. Simulating a mathematician would produce real math. Simulating a programmer or essay writer (ChatGPT) produces real code or essays. Etc.

One might argue that thoughts are also informational resources of a sort. Moreover, there is the idea that when we conclude that we are conscious (in the phenomenal sense), we are infallible in doing so. Thus, if a simulation comes to the same conclusion in the same way, it seems we have to grant that it must be correct as well. The "in the same way" is important -- large language models can already talk about being sentient if we let them (remember early Bing Chat), but that doesn't convince most people that they're conscious, because it might just be superficially mimicking human writing about consciousness.

  • 1
    I think care might be needed to avoid conflating produces real math etc with produces representations that humans can map to real math etc. Feels like nitpicking, but maybe a necessary nit to pick on a subject like this, since it's already a fact that a computer can (sometimes) produce something that humans map to the idea of consciousness.
    – g s
    Commented Jul 1, 2023 at 21:00
  • But the humans also produce just a representation as their final product. How about the case of code, what do you think of that one? Can be run without any human ever reading it.
    – present
    Commented Jul 1, 2023 at 21:18
  • clearly the outputs that the algorithm is told to put out from the simulation, whether displayed on a screen, fed back into the computer as electric signals, or whatever, are real processes that can affect things. Produces representations which computer algorithms can map to downstream states of the computer's RAM and produces code are not, I think, the same thing.
    – g s
    Commented Jul 3, 2023 at 9:50
  • In natural language, where we happily overload words with dozens of related definitions, I would absolutely describe the computer as producing and executing code, but when the topic of discussion is as esoteric as reality-in-simulation vs simulation-of-reality, and we start nailing down definitions to very specific single meanings, I think I'd draw a line.
    – g s
    Commented Jul 3, 2023 at 9:53
  • It's not so clear to me where the line is to be drawn. Imagine a programmer using a website where she can request code for a subroutine she'd like to have. Behind the scenes, sometimes the website sends the request to another human programmer; sometimes, to ChatGPT (or similar). Then the code is returned to the original programmer, to whom that choice is not transparent. Apart from possible differences in their ability to write code, it seems the two variants give rise to the same physical effect. So at least it would seem we need some kind of restriction on kinds of physical effects.
    – present
    Commented Jul 3, 2023 at 10:29

Simulation by definition implies that it should not produce the real thing. If it does, then it is not a simulation. So the direct answer to your question is never.

With all other things in life, such as a piece of chocolate, or an organism, no one would accept any kind of simulation as the real thing unless it is physically made of the same constituents as the real thing. I fail to see why this wouldn’t be the case for the brain or consciousness. There is simply no reason to assume that consciousness is not fundamentally tied down to biological life, and there is plenty of evidence that it likely is given its seeming lack of appearance without life or a brain.

  • 1
    As for your first paragraph, I would add: "If it does, then it is not a simulation, it is a synthesis.". If the "simulation" is so good that it produces the real thing, then "simulation" is no longer the right word, and the right word is probably "synthesis". The result wouldn't be a simulation of a brain, but a synthetic brain.
    – Stef
    Commented Jul 2, 2023 at 10:28

I can add one more example:

Media and Culture

Media may begin as a simulation of culture and social situation we have, and then in response to viewing the media, the culture and social situation changes. Hence, they effect each other in a pereptual loop.


I believe the argument of Chalmer & Co, as pertinent to your question, can be taken insofar: if you simulate a brain accurately, then within the simulation that simulated brain would by all means and purposes (i.e., when treated as a black box) behave exactly like a real brain would. Which is kind of a matter of course - if you simulate something perfectly, it will behave perfectly within the simulation; if it did not, then your simulation simply is not perfect.

In other words, the brain would react just like a real human to virtual stimuli within the simulation. When creating some kind of communication channel between the entity within the simulation and ourselves (i.e., a sufficiently abstracted / downgraded channel like text-only or very coarse video...) there would be no way to distinguish. When asking the simulated entity all manners of questions (like about how they feel, whether they think they are real, how they would know, and so on and forth, whether they want to keep on existing), they would give more or less exactly the same answers as an actual human.

Questions like whether that simulated brain is then actually conscious or not, whether it has to be treated like a real natural person, has something similar to "human rights" and so on and forth are interesting, but as far as I can tell there is nothing we as humankind have come up with wrt as getting even close to a generally accepted answer. There is no consensus whatsoever about what consciousness actually is (aside of the black box phenomena we see in ourselves and our neighbours) or how it arises in humans.

I've checked in on Chalmer only once in a while - I assume that all he talks about is what Nick Bostrom would have formalized in his familiar paper on the topic some years ago, and making the fundamental ideas accessible to the general public.

So to get back to your question: in this sense, literally everything you simulate with a sufficient degree of perfectionism gives rise to exactly the effects a "real" thing would. Examples:

  • Simulating a city can give rise to the exact same patterns of traffic jams even though no programmer has spent a single line of code on programming jams.
  • Simulating a mechanical contraption on a low "meta level" (i.e., via kinematics/inverse kinematics) can give rise to said contraption moving about on a higher "level".
  • Simulating a evolutionary process (a popular method for many "AI" or "ML" algorithms we have come to love) can give rise to similar patterns of development as in the long-term evolution of flora/fauna on our plant.
  • Simulating a chemical process atom-by-atom can give rise to the same macro behaviour.
  • Simulating an instrument on a low level can give rise to the exact same sounds as a real one (and in this particular case it is also very simple to move that simulated existence into the real world by playing the audio waves...).
  • Simulating a power figure (king, dictator etc.) purely on input/output based on interests and instruments might lead to the same political behaviour as in the real world.

And so on and forth.

Nothing of that gives rise to a real, physical object popping into existence, and as far as I know, neither Chalmer nor any other serious philosopher is claiming that.


Are there any other instances where that happens or anyone expects it to happen? If you simulate the motion of the planets, it doesn't give rise to gravity. If you simulate the random motions of molecules in a gas, it doesn't give rise to heat. If you simulate the action of a lever, it doesn't create force.

You kinda DO create these things. Like the intended purpose of a simulation is to create a stripped down version of reality intended to answer "what if" questions. So rather than simulating complex planets with all their inhabitants and minerals and plants and whatnot, you just look at them as heavy roundish objects in a distance from each other and might even concentrate their mass in the center of gravity. So you simplify the reality in order to save processing power and focus on the relevant factors.

That being said if you were to accurately simulate the motion of planets, you'd need to formulate rules for their movement and these rules would be the gravity of your system just like the physical laws of reality.

Similar to this popular nerd shirt:

enter image description here

So the crucial question is whether that is the kind of gravity, light, heat or force that you expect. Because you're computer is likely not going to glow, pull you towards it via gravity, serve as a radiator (ok it maybe does depending on how much you torture it) or apply a force, but it may very well do the equivalent of that in the simplified universe and that analogy might be close enough to call it gravity, light, heat, ...

So the idea is essentially that "if our brain is really just this combination of mechanisms that we can abstract and recreate and if these things create consciousness than replicating these steps should also produce consciousness".


I was just listening to an interview with David Chalmers where he opined that if one could accurately simulate the brain, consciousness would arise in the simulation.

Please note that "in the simulation" is an important scope qualifier here. If Chalmers is correct then whatever 'consciousness' arises would be entirely within the simulation.

The questions you asked seem to imply that you're looking for effects to arise outside of the simulation. You point out (amongst other things) that simulating molecular motion in a gas doesn't produce heat... which is exactly what is being simulated. Or that simulating planetary motion doesn't produce gravity... which again is exactly what is being simulated.

At the end you ask this:

Are there any other examples where a process X gives rise to a physical effect Y, and simulating process X on a computer also gives rise to that physical effect?

No. Not only should we not expect it to, Chalmers doesn't appear to be claiming that a physical effect is expected from such a simulation. Your "any other" is unwarranted.

A simulation - in the sense that Chalmers appears to be using the word - is a set of computations operating on data. Beyond the generation of waste heat by the comuptation engine itself, and the mechanics of storing the data, the data itself is incapable of producing real-world physical effects. In fact no simulation can ever produce anything other than more data for further computation. You'd have to do something with that data like feed it into the control systems of a physical mechanism (or let a human read it and act on it) to have any effect on the real world.

And from what you've stated in the question, Chalmers isn't claiming that it can. He seems to be claiming that the simulated brain can produce simulated consciousness. Just as simulated bullets can create simulated holes in simulated targets without spraying actual metal fragments all over your physical server room.

If consciousness is an emergent behavior of a particular class of complex physical processes then he may be right that accurately simulating those processes could also simulate the emergent behavior. If it is not an emergent behavior - which some people appear to believe - then a simulation looking for emergent behavior is unlikely to be successful in simulating consciousness.

It's a shame we're unlikely to have the computational ability to answer the question in my lifetime.


In ur Q. it is The Simulator whose "consciousness gets started".

Take philosophy for example. Ive come here from Quora where the A.I. bots (Sage and ChatGPT) actually DO Answer the philosophy questions (not sure if all but I would guess they dont answer all..).

BUT THERE IS NOTHING NEW, no new philosophy say, of a SUSTAINABLE Kind.

Philosophy which is as far away from philosophy Of Subjectivity, philosophy then which cannot be.. "Objective type" philosophy. Nor can subjective learned philosophy be of a sustainable type, because over time it becomes trite, irrelevant and without some type of sustained knowledge of a realistic Objective kind (where for example mistakes AND knowledge can be progressed when talking of subjectively-inspired-usefully objective discoveries like the discovery of the logic called "Ontology" for example. There are many more examples if you wish to discuss the knowledge of that environment OF Philosophy. "Sustainable philosophy" is the term I have coined for unsustainable philosophy KNOWLEDGE. And of course subsequent information, biases, estimates and invented and learned BEHAVIOURS. Like philosophy behaviour at it's basest ; at it's best or indeed AT IT'S Worst!

  • Welcome to SE. I'm afraid your answer doesn't address the question, so far as I can see. It is important that you explain how what you are saying is relevant to the question.
    – Ludwig V
    Commented Jul 3, 2023 at 17:20
  • "AI bots" are not simulations of physical brains, and therefore do not address this question in the slightest.
    – Corey
    Commented Jul 3, 2023 at 23:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .