1

I was wondering what kind of computational model is the human brain (as it seems superior to a Turing machine).

Another thing that should be a separate question, What would be a perfect computer model be like, is our brain any close to it?

Apology for putting the ideas in a very beginner way, which ofcourse I am.

Aether
  • 133
  • 3
  • Is this answered already by https://cstheory.stackexchange.com/q/5379/129? On a related note: why do you think the brain can solve problems that a TM can't? – Joshua Grochow Jun 26 '20 at 04:06
  • @JoshuaGrochow like there's not a good algorithm that can make any algorithm. Maybe I am wrong. Please correct me. – Aether Jun 26 '20 at 04:18
  • In my understanding of the brain, one can model the brain quite faithfully on a (albeit very large) Turing machine, by simulating all the molecules in the brain if it came to that. So in terms of the brain's ability to solve computational problems, I think there are always TMs that, in principle, could do the same. – Joshua Grochow Jun 26 '20 at 04:25
  • Define "brain".... – Avi Tal Jun 26 '20 at 04:28
  • @JoshuaGrochow apologies for the followup beginner question. Is what you said known for a fact? (I mean, the fact that we can simulate a brain with a TM). For instance, what about quantum interactions between particles? – olinarr Jun 26 '20 at 06:46
  • @olinarr Nothing is known for a fact in natural sciences. The best you can do is to formulate hypotheses consistent with the available observations. – Emil Jeřábek Jun 26 '20 at 09:38
  • 2
    @olinarr: While Emil is right that nothing is known for a fact, even for quantum interactions, sure you can simulate them. It might take a long (exponential) time (closely related to the BQP vs P question), but certainly a TM can do it. I don't think there have been any physical models proposed (general relativity, quantum field theory, etc.) where simulating them is uncomputable. – Joshua Grochow Jun 26 '20 at 18:05
  • The Church-Turing hypothesis is that Turing machines model what can be computed in this universe. Arguably, everything brains do is computation in this sense, so Turing machines are powerful enough to (theoretically) model what brains do. (Of course this doesn't mean that existing computers are as powerful as brains.) If brains compute in some manner that TM's cannot model, then, since brains are subject to the laws of physics, the Church-Turing hypothesis would not hold, which would suggest the possibility of some more powerful model for what is physically computable. – Neal Young Jun 26 '20 at 19:52

1 Answers1

2

This is pretty much an open problem and subject to active research. There are a few proposals available. Here are some of the latest ones:

  1. Brain computation by assemblies of neurons Christos H. Papadimitriou, Santosh S. Vempala, Daniel Mitropolsky, Michael Collins, Wolfgang Maass Proceedings of the National Academy of Sciences Jun 2020, 117 (25) 14464-14472; DOI: 10.1073/pnas.2001893117 [Link]

  2. Manuel Blum's model of consciousness: [Video]

Mahdi Cheraghchi
  • 4,031
  • 22
  • 30