The brain in a bad vat argument states that we may be a brain plugged up to a supercomputer that simulates the external world by sending signals to the brain, just like the brain normally receives signals. The simulation argument states all of reality is in a supercomputer. But, what is the difference between the two?

In the brain in the vat argument, the brain would still have to be simulated too. Otherwise, in this simulated world, you are a body without a brain (and your actual brain is outside the simulation). Also, a body walking around without a brain would violate the physics of the simulation anyways, mainly conservation of energy. So, in both scenarios, the whole world is simulated, except in one scenario, there is also an external brain. Is there really a difference? Is there really a need for an external brain if you're going to have to simulate the brain anyway?

  • Also compare Descartes' "evil genius" and Edgar Allen Poe's "All that we see or seem. Is but a dream within a dream." There is a common thread to all these which give rise to Solipsism: essentially we cannot know anything else to the same degree of certainty as that we ourselves exist.
    – christo183
    Aug 5, 2019 at 15:53
  • 2
    Brain in a vat still assumes existence of the brain, as something not simulated. I.e. your senses deceive you, but your mind is still "natural" . Simulation goes step further, now even your consciousness and your "free will" is simulated .
    – rs.29
    Aug 5, 2019 at 18:25
  • solid question, thanks
    – user38026
    Aug 8, 2019 at 18:06
  • Consider the brain-in-a-vat thought experiment, and the idea of a simulated reality as, to use a computer analogy, upgrades made to the age-old Cartesian deus deceptor (evil demon) scenario. Remember the 1996 movie adaptation of Shakespeare's 1597 play Romeo & Juliet? Swords were replaced with guns. Reminds me of Sphinx riddles somehow. Feb 23, 2022 at 13:13

6 Answers 6


One difference between the brain in a vat argument and the simulation argument is their motivations. The brain in a vat argument is motivated to support skepticism through doubting one's senses. The simulation argument on the other hand expects one to actually believe that is how reality works.

Here is Wikipedia's description of the simulation hypothesis:

The simulation hypothesis or simulation theory proposes that all of reality, including the Earth and the universe, is in fact an artificial simulation, most likely a computer simulation. [my emphasis]

Note again my emphasis on "in fact". This is not an attempt to argue for a skeptical approach to our perceptions of reality, but to argue in favor of one particular explanation of those perceptions.

Here is Lance P. Hickey's description of the brain in a vat argument:

The Brain in a Vat thought-experiment is most commonly used to illustrate global or Cartesian skepticism. You are told to imagine the possibility that at this very moment you are actually a brain hooked up to a sophisticated computer program that can perfectly simulate experiences of the outside world. Here is the skeptical argument. If you cannot now be sure that you are not a brain in a vat, then you cannot rule out the possibility that all of your beliefs about the external world are false.

This argument encourages one to doubt any explanation one might have based on our perceptions of reality. If one wanted to be skeptical about the brain in a vat itself one could always replace the vat with Descartes' demon deceiving one about the existence of the vat. This argument is not trying to convince one that there actually is, in fact, a vat with one's brain inside it.

Hickey, L. P. The Brain in a Vat Argument. Retrieved on August 5, 2019, from Internet Encyclopedia of Philosophy at https://www.iep.utm.edu/brainvat/

Wikipedia contributors. (2019, July 18). Simulation hypothesis. In Wikipedia, The Free Encyclopedia. Retrieved 16:09, August 5, 2019, from https://en.wikipedia.org/w/index.php?title=Simulation_hypothesis&oldid=906844036

  • Wiki: "all of reality". You: "skeptical approach to reality". You and Wiki make a specious use of the word "reality". Reality is whatever exists and I don't know of any argument concluding that reality doesn't exist.The Brain-in-a-vat argument is scepticism not about reality but about our perception of the physical world. Similarly, the idea of a simulation doesn't question the reality of reality. How would that work?! It questions our naive belief in the reality of nature, or in the reality of the natural world. In both cases, reality is safe. Aug 5, 2019 at 18:33
  • @Speakpigeon I made an edit hoping clarifying that the skepticism is about our perceptions of reality. Aug 5, 2019 at 18:46

There is indeed a fundamental difference.

The Brain-in-a-vat seems within the reach of even human technology. In other words, it is a highly convincing and realistic scenario. A brain in a vat would presumably take the simulated physical world to be the actual world.

The idea of a simulation, that the world itself, including the brain experiencing this world, could be a simulation is much more fantastic and therefore somewhat more difficult to accept as a possibility.

The main sticking point, however, is that we still have no explanation as to how our subjective experience could possibly be a property or consequence of the way our brain works. The idea of a simulation requires that we accept the idea that our subjective experience could possibly be a creation of the simulation and therefore, fundamentally, an illusion (see note).

In the Brain-in-a-vat, the nature, or indeed natures, of both our brain and our subjective experience remain exactly as each of us happens to believe that they are. With the idea of a simulation, both our subjective experience and the physical world are turned into illusions.

The Brain-in-a-vat, although more realistic and conceivable, is however entirely metaphorical in its motivation. It does not even try to suggest that you really are a brain in a vat. Rather, it is an argument, a logical argument, to help us understand why we cannot be certain of the reality of our perception of the physical world and, hence, of the reality of the physical world itself as we think of it.

The idea of the simulation is not an argument. It is a metaphysical claim about reality. A such, it is to be seen as connected with the idea that consciousness is a process and, therefore, to the idea that computers can become conscious. All that would be required would be that the software got to a sort of critical threshold of complexity.

The Brain-in-a-vat, on the contrary, suggests a decisive epistemological dualism between our own mind that we know, and that we therefore know that it exists (Descartes' "I think, therefore I am"), and the material world that we can only believe in, and that therefore we don't know that it really exists.

In effect, these two ideas are polar opposite. The Brain-in-a-vat says the physical world may not exist, while the simulation says that our subjective experience may be just an illusion.

To the extent that they are polar opposite, I don't see how anyone could see these two ideas as equally convincing. If you find them equally convincing, it is likely because you haven't understood at least one of the two.


The difficulty in believing in the possibility of our own consciousness being a simulation is that the only notion we have of what a simulation is comes with our notion of a material universe: we can only conceive of such a simulation as something of the same nature as that of a material universe. This is problematic for two reasons.

First, if the simulation was material in the same sense as that our material world, then we would be part of the material world of the simulation, which is definitely not what the idea of a simulation is meant to suggest to begin with. If instead the simulation was material in some different sense as that of our material world, then the notion of simulation becomes meaningless simply because nobody will be able to explain in what sense it would be material.

The second reason that it is problematic to believe in the possibility of a simulation of our consciousness comes from the fact that we are still unable today to explain the subjective quality of consciousness in terms on the material world.

A simulation, to the extent that we are only able to conceive of it as something akin to the material world, would be just as useless in explaining the subjective quality of consciousness. Merely assuming our material universe is a simulation fails to explain the facts of our consciousness.

In other words, the idea of a simulation doesn't make sense and doesn't explain anything.

  • Although it's fair to say that these hypotheses had different reasons to be introduced, you seem very confused about their consequences. If either were true, you would have no direct evidence about the matter in which consciousness resides, so you would have almost no reason to believe a brain in a vat would be more plausible than pure simulation. In fact, Occam's Razor would compel you to believe it's less plausible.
    – Veedrac
    Aug 8, 2019 at 6:22

Both arguments are skeptical about the "real" existence of the external world as we believe it to be, or seem to experience it, but there are some significant differences.

The Brain in a Vat doesn't make any particular claims about the nature of consciousness, or the existence or non-existence of the soul, except for the assumptions that both consciousness and the soul, if they exist, are seated in the brain. We are arguably close to achieving a level of technology sufficient to put a brain in a vat --it would require only a) that we could wire the brain up in such a way as to replace the sensory inputs coming from the sense organs with simulated inputs and b) that we could generate convincing sensory inputs. This second requirement might seem to entail simulating an entire universe on a computer but it doesn't require simulating consciousness, and it doesn't require an entire simulated universe (just the portion the brain thinks it's interacting with). Except for the direct wiring to the brain --which already is being used in a limited form for people with sensory handicaps --, it's not terribly distant from contemporary "virtual reality" technology. The point of the "brain in the vat" thought experiment is to encourage us to accept that some or all of our beliefs about the external world might be mistaken.

The Simulation Argument assumes that everything in the universe, including consciousness, is capable of being reduced to physical/physicalist processes, and further, that such processes are capable of being simulated on a computer. As formulated by Nick Bostrom, it attempts to show that this it is not only possible, but also significantly probable that what we experience as the universe exists only on a computer.


More of the same

There is little practical difference among the different scenarios. As christo183 points out, there have been many variations of the same thought experiment, and they all basically have the same point – How do we know what we know about the world?

Different implementations of the scenario were thought to be more or less plausible when they were described. The ancients pondered the difference between being awake or dreaming. In another time, demons might have been thought to be real. Some people still think they're real. With materialism and the loss of mind-body duality, brains in vats seemed more plausible. Now with neural networks and genetic algorithms, simulated realities seem reasonable to consider. It's a progression, updated for the times.

Each update brings with it new questions. – Is the "consciousness" of a simulated brain the same as the consciousness of a "real" brain? – But it's really more of the same. How can we know what is real?


With simulations, there can be recursion. Simulations running in simulations. Since layers upon layers of simulations can be embedded within each other ad infinitum, some people believe that makes it highly probable that we are already in a simulation. But they reach that conclusion by hand waving the math. No one passes math class that way. No one can figure out what's likely or not that way either.

Suppose there is a 1/10 chance we are in a first-level simulation vs the "real" world. The second level simulation would require resources equivalent to that needed for the first-level simulation, 1/10 (of 1/10). Similarly for the third level, 1/10 (of 1/10 of 1/10), and so on. If we add all these probabilities up, we get 1/10 + 1/100 + 1/1000 + 1/10000 + ... = 1/9. The odds of being in any simulation, even if there are an infinite number of them, vs the real world is 1:8.

I am assuming a geometric series of probabilities. Perhaps that's not how it works. Whatever the probabilities are, they're not constant or increasing, since that would cause the sum to rapidly exceed 1, which is impossible. The point is the sum of an infinite series may be a rather small finite number. This is a counter example to the claim that the probability of being in a simulation must be high.

Virtual Machines

Consider computers, such as the one you're using to read this. There are emulators and virtual machines, which are basically simulated computers running on other computers. If I run a program, is it highly probable that the program is running on a virtual machine vs a real machine? Suppose it's confirmed to be a virtual machine. How many layers deep is it likely to be?

There are limitations to running multiple layers of virtual machines:

  • Hardware limitations. The virtual machine requires hardware support that is not supported by other virtual machines. Real, physical hardware is required.

  • Utility limitations. There isn't much point running simulations upon simulations.

  • Computing limitations. Processing cycles have to be shared with the host layer.

  • Resource limitations (memory, storage). Each layer can only use a small portion of what was available to the host layer. 

Recursion has limits. We may not know exactly what they are for simulated realities, but it doesn't make any sense to ignore their existence to claim unreasonable results are highly probable. While hardware and utility limitations may not apply to reality simulations, computing and resource limitations are likely.

Vat vs Simulation

Some claim that we're "closer" to achieving a brain in a vat than we are simulated realities. Given the expansive nature of simulating all of reality, I'd agree. (Argually, the universe is the most efficient universe simulator. At best, we could simulate only portions of reality.)

But at most, all any individual has evidence of is a single simulation of consciousness. In that sense, we are closer to simulating consciousness than we are to keeping brains in vats.

To simulate consciousness, we need only replicate the neural network. We don't even have to design it. Just copy it from an existing brain. We'd have no better idea of how or why it's conscious than we do our own brains, but it's doable with sufficient computing resources.

On the other hand, keeping a brain functioning in a vat would require not only knowing how to keep it alive, but also how it functions so that it could be appropriately stimulated. It's "within the reach of even human technology" the same way Krang is. 

  • (1) Calculating recursion probability assumes existence of finite sized first level. If size is infinite or levels are cyclic, however small the successive probability factor, we are likely in simulation. (2) Virtual machines need not be less efficient than native. A better analogue will be abstractions e.g. C vs assembly programming language. Beings in assembly language are simulating a constrained subset called C world where operations are as fast as equivalent assembly instructions (though kinds of operations are constrained). Both are interfacing with same substrate (CPU).
    – tejasvi88
    Feb 22, 2022 at 11:55

Both hypothesii are that we have can have no direct evidence that what we perceive is not a convincing hallucination; because any such evidence would also be "what we perceive".

The "brain in a vat" hypothesis assumes that there is something doing the perceiving, it just need not be where it is perceived to be.

The "simulation" hypothesis just takes that a step further and acknowledges that it need not even be what it is perceived to be.

The simulation hypothesis suggests we can not even know the nature of the simulator, because any evidence about it would also be part of "what we perceive". The best we can do is examine what we perceive to determine if it is plausible that it might be produced by a simulation.


The modal markers are very, very important. There is a may in the brain in a vat theory. There is an almost certainly in most versions of the simulation hypothesis. (In your own statement of it, the uncertainty seems to have totally disappeared. The real argument does not take us that far.)

The brain in a vat argument suggests that we can't tell whether or not we are viewing reality or a simulation. It does not weigh in on the likelihood. In fact it suggests that we couldn't get a handle on the likelihood if we wanted to, because we can't get any data that is both clearly about something and couldn't be manipulated by the simulation's author.

The simulation argument layers more upon that: it looks at the recursive application of this possibility and by a sort of infinite Baysian inference decides that it is highly likely that we are not living in realty, but in a simulation.

So the second of these builds upon the first one and uses a statistical view of either motivation or effective action to discuss the uncertainty involved, and reduce it.

You must log in to answer this question.

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