This problem is not the idea of storing the brain on a computer, which I believe could be possible one day (though I think the ethics of doing so could be not too terribly great). That is a completely different problem than the concept of simulating a whole brain on a computational model.
Let me use an analogy to set up the problem. Suppose I were to build a bridge. This bridge is to extend from Chicago, to Ludington (across Lake Michigan). The engineers building this must do so with testing the foundation in pieces. Do you know it will actually work by just building the bridge as a whole? Not at all. The composition of a bridge's parts doesn't mean the bridge itself can sustain the capacity.
Often I find this idea is something that excites primarily people who are not exactly computationally literate in theoretical computer science. If they maybe took up a little looking up of the past in our field, they would know that this concept is a little off the wall silly to carry out with the amounts of funding (millions upon millions) pooled in already.
Since 1936 when we developed the last of the several nails in the limitations of Logic, we learnt that as humans, logic runs on the rules given by the human. We would say logically that something is true if the premises are true, but automatons don't exactly view things that way. You cannot give every assumption that brain can handle due to the concept of undecidable problems. Humans believe it or not deal with these types of problems on a near daily basis. Computers can't handle paradoxes effectively if given arbitrarily. There is bound to be a new form it must handle, and such is silly to think we can solve that. Our brains could definitely be mapped, scanned, stored, but to use a computer to carry this out means we have a lot of problems in computation which cannot be resolved due to undecidable problems. Such problems are the Halting Problem, the Mortality Problem (especially), and necessarily, infinitely more to deal with (since you can never cure incompleteness).
In conclusion, to assume we can bypass undecidability is without a question a ridiculous idea until we have:
- An algorithm which can indeed do this (which they have not demonstrated, since this brain must be able to do that)
- A proof of correctness for this algorithm
The reason why this is such an issue is because if one did this on a computer, it would contradict Godel's 2nd Incompleteness Theorem which is a fundamental result which limits the mathematics we can prove with theorems.
In basic, you need to show such is even possible with computation before setting out claiming such is possible. It is a very fine line.
In basic, you need to show such is even possible with computation before setting out claiming such is possible. It is a very fine line.
Have a beautiful day!
D R Page
I don't understand your invocation of Gödel's (2nd) incompleteness theorem. Would you please elaborate?
ReplyDeleteThere are limits in computation, regardless of the model. You could employ either the Church-Turing theorem, or Godel's 2nd incompleteness theorem to say this with authority. It all boils down to how we can decide on problems. Nobody has proposed an algorithm, nor a proof of correctness for such algorithm to contradict this notion yet, but still these quacks insist it is possible (yet), yet haven't provided anything in theoretical computer science to go right ahead to do this work. We have not seen the model which trumps any variation of the Universal Turing Machine yet. A responsibility of a researcher is to prove something is even possible in that realm before proceeding. Like if I know a certain construction exists, I will proceed to find out its decidability before even considering the designing of an algorithm.
ReplyDeleteThis is one of my research areas.
Hope this helps!
D.R. Page