I was wondering about these skeptical hypotheses regarding the reality we experience. More specifically I wanted to address the Simulation and Brain in a Vat arguments because to me they seem kind of similar (correct me if I am wrong). In both cases there should be some sort of computer simulating the reality I perceive. A simulation is a representation of something else, right? My question is:

Is it possible to create a perfect representation of something else?

If I see an apple in the actual reality I would notice the difference from an apple on the screen of a computer or on the canvas of a painting. Even the most photorealistic VR could never give me the same experience of an apple as if it was real. I say this because:

  • My consciousness is certain to exist;
  • My consciousness is part of the actual reality even if what I perceive is not;
  • My consciousness is equal to the consciousness of the beings who live in the actual reality;
  • Therefore I would be aware of something less accurate than what they perceive.

It's like being born in a super photorealistic virtual reality without noticing the slight inaccuracies of what I sense. I mean after all I am as conscious as the aliens who live in the base reality, right? So they would notice the differences...but not me?!

Someone could argue that it's like Plato's allegory of the cave, but perceptually speaking the shadows projected on the two-dimensional wall of the cave are as accurate as the shadows that exist in the three-dimensional reality. So I don't think that's a good argument.

  • Sorry, but this is not a good argument. You are elevating your personal impressions to what is possible in principle, this is the kind of reasoning that led ancient humans to believe that the Earth is flat, etc. One has to make corrections for our limited experience and lack of imagination. Also, as neuroscience and psychology experiments showed in recent decades, "consciousness" is a highly unreliable witness to anything: people routinely report forming an intention to move a finger after their muscles already start contracting. And no one is even trying to trick them.
    – Conifold
    Aug 24, 2018 at 0:12
  • 1
    @Conifold "consciousness is a highly unreliable witness to anything: people routinely report forming an intention to move a finger after their muscles already start contracting. And no one is even trying to trick them". Behaviour as a measure of consciousness is UNRELIABLE. Neuroscientist have no way of telling if there is consciousness before behaviour. So I wouldn't put forward the poor science of consciousness as an reasonable argument. "this is the kind of reasoning that led ancient humans to believe that the Earth is flat". Absolutely not. I am simply stating what a simulation is.
    – user34482
    Aug 24, 2018 at 0:36
  • 1
    By the way, to clear things out, you're saying that consciousness is unreliable because neuroscientists have made discoveries on the time passing from an unconscious to a conscious state of an action. I am not talking about timing or anything like that. I am talking about the conscious state, when we are aware.
    – user34482
    Aug 24, 2018 at 1:10
  • 2
    I would use Searle's Chinese Room Argument to claim that we are not a simulation nor are we brains in a vat nor will AI ever be able to understand anything. Aug 24, 2018 at 1:27
  • 1
    It seems that you are not familiar with the background of these experiments or with the psychological literature on confabulation (making up of intentions after the act), see a recent survey by Roskies. Even in the so-called conscious states much of mental activity remains unconscious, and in any case, one certainly can not expect to be aware whether a conscious experience is externally induced, think of hallucinogens. The burden of proof is on you to show that introspection is reliable, otherwise your argument is circular.
    – Conifold
    Aug 25, 2018 at 3:25

1 Answer 1


This is a comment on the idea that a person, e.g. me, could really be a simulation running on a computer.

The idea that I'm a simulation assumes that the executing simulation has semantic content. I'm nothing more than a simulation, but nevertheless I understand things, enjoy (virtual) KFC, think its bad for me (but it isn't because I'm just a simulation), perceive the world, etc., (have a semantics), but in fact I'm just a program running in a computer.

So the simulation that is me has a semantics, even if erroneous. The problem is – could a simulation have a semantics? One reply is: It depends what you mean by simulation. But happily, Turing (in 1936) gives an answer. A simulation is a description of a machine (or natural system). In other words, it's a thing made of symbols that have meanings (to the cognate entity writing the description). And a computer or Turing machine runs the description.

While doing this, the computer also receives input symbols, which describe an input that the machine being simulated could receive. The computer then emits output, which output symbols describe the output that the machine being simulated would have emitted had that machine or natural system actually been running and received the input described by the input symbols. E.g., simulation of a weather system.

So the question, Where is the semantic content in a simulation? becomes: Where is the semantic content in an executing computer program? The answer: in the brain of the programmer who wrote the program (the description).

Simulations contain lots of conditionals, since these are needed to take account of the different possible inputs that could arrive. And the conditionals have the general written-down form: If INPUT = "A" then OUTPUT = "B". But there are no semantics in these conditional. The semantics was in the head of the programmer. The programmer knew that output symbol B describes an appropriate response of the machine being simulated to the input described by input symbol A. That's why the programmer wrote: If A then B, rather than C, D, E ... What the symbols A and B mean was in the head of the programmer. But the only thing in the program is: If INPUT = "A" then OUTPUT = "B".

Hence a simulation lacks semantic content. And if it has no semantics then it cannot be me, since I understand stuff, and having semantic content is necessary to my understanding stuff. (Which is just a version of the Chinese room argument.)

You must log in to answer this question.