Quantum Mechanics and Consciousness

<< < (4/12) > >>

Peter Grant (September 08, 2009, 20:50:58 PM):
Thanks for the lengthy response, quite illuminating. :)

I wouldn't argue that human brains don't have a tendency to sometimes spit out garbage…
True enough, but quantum computing is more akin to individual neurons occasionally firing haphazardly, rather than the whole brain going off the rails. As long as some majority (above a certain threshold) of the neurons fire in concert, the brain will produce a definite answer (or so neuroscientists are currently inclined to believe). The correctness of the answer will depend on several other factors which can be loosely grouped under the heading “inputs.”


OK, so in techie terms it's a bit like error handling through redundancy?

Quote from: Mefiante
… but how quickly and accurately do we actually solve these BQP problems?
Theoretically, a quantum computer would solve certain general instances (NB! not all of them) of BQP (and P) problems very, very fast – much faster than even the most powerful modern supercomputers and superclusters that presently are capable of speeds in the order of ten teraflops. However, there are some tricky technical problems to overcome first before quantum computing becomes a reality. Execution time on a quantum computer will grow sub-linearly with desired accuracy.


Makes sense.

Quote from: Mefiante
Wouldn't ordinary deterministic computing provide sufficient explanation for our somewhat limited abilities?
It seems unlikely. If it was so, we would probably be able to simulate a human brain fairly decently, which is something we still cannot do. Moreover, the brain can exhibit problem solving behaviour that is outside the P complexity class. For example, often when you solve a crossword and are looking for the answer to a particularly vexing clue, you don’t systematically run through all the possibilities, although naturally you’ll try a few. The answer, when it comes, suddenly pops clearly into your head and is almost instantly recognised.


Not sure I understand this P complexity thing, but the crossword example doesn't seem all that complex when compared with problems like working out prime factors for instance. Why could a clever subliminal search algorithm, or an exceptionally ingenious indexing system not account for this?

Quote from: Mefiante
Does proposing a quantum brain offer any explanatory value?
Not at present but it may do one day, although it does seem rather unlikely. Consciousness is an essential dimension of a functioning human brain but nobody knows what consciousness is. The squabble in this thread is over just this crucial issue: the complete QM-consciousness model is still a wild guess because it is missing several key ingredients. It is scientific only insofar that some aspects of it can be tested, at least theoretically, but the essential difficulty of how QM effects produce brain activity (and/or consciousness) remains entirely obscure and a matter of considerable speculation. Also, a large part of the problem is that “quantum” has in many quarters (especially in New Age ones) become the next Supremely Transcendent Universal Principle of Ignorant Deduction (STUPID), much like god/gods was/were in the past: “I don’t quite know how this works, so it must be quantum. Hallelujah!”

'Luthon64


Yeah, this worries me too. Assuming for a moment we and most other vertebrates, as I think one would have to assume, have evolved this remarkable quantum computing ability, what does it have to do with consciousness? If these quantum effects are taking place, they are necessarily subliminal, occurring in tiny quantum intervals far too brief for us to be consciously aware of.
Mefiante (September 09, 2009, 10:40:43 AM):
OK, so in techie terms it's a bit like error handling through redundancy?
Er, not quite. It’s rather more like a Monte Carlo simulation where the solution emerges from an aggregate of a large number of statistically random trials or “samples” from the problem space. What happens in a sense is that individual errors are random but they tend to cancel one other out in the long run, and said aggregate converges on the required solution. Just as in ordinary statistical sampling, the larger the number of trials, the more confident one can be that it is properly representative of the whole.



Not sure I understand this P complexity thing, but the crossword example doesn't seem all that complex when compared with problems like working out prime factors for instance. Why could a clever subliminal search algorithm, or an exceptionally ingenious indexing system not account for this?
The point though is that from the brain’s perspective, solving a tricky crossword clue is procedural (or algorithmic) only in a very loose sense. As said, you’ll try a few solutions before suddenly hitting on and recognising the correct one. Speaking strictly algorithmically, the process would be quite different. It might involve a base dictionary of all possible words, selecting the correct (or probable) language, counting letters, extracting a subset of candidate words based on their length and known letter positions, and so on. Finally, the correctness or otherwise would be judged by assessing the crossword puzzle as a whole, and the process may actually find more than one solution.

The only essential difference between solving a crossword in this way and factoring a large composite number is in the size of the solution space. The totality of human words consists perhaps of a few million, whereas factoring a 100-digit composite number has a solution space of around 1050 (a one, followed by 50 zeroes) possibilities. Both problems actually sit in NP complexity space because the solution space grows exponentially with problem size. Also, solving a crossword does involve some form of subliminal search algorithm. The question, however, is whether that algorithm is deterministic or probabilistic. There is good reason, as outlined earlier, to think that it is the latter, and quantum computers would also run algorithms. The use of an algorithm does not mean that a solution process is necessarily deterministic.

Concerning an indexing system, this itself needs to procedural/deterministic otherwise it would at times fail to index the same thing consistently. In the context of the crossword problem, it seems obvious that the indexing would need somehow to index the meaning of words, phrases and sentences (and we’ll even ignore crosswords that provide cryptic clues only). But meaning is notoriously difficult to pin down in many cases, and furthermore the process of extracting such meaning would itself need to be deterministic-algorithmic for indexing to work correctly. Yet the brain appears to act associatively (as opposed to procedurally), often through pattern-matching on templates that can be somewhat fuzzy or fluid.



If these quantum effects are taking place [in the brain], they are necessarily subliminal, occurring in tiny quantum intervals far too brief for us to be consciously aware of.
Indeed, and that is one of the major objections against even just the concept of a QM-mediated model of consciousness.

'Luthon64
Peter Grant (September 09, 2009, 15:41:42 PM):
Thanks Mefiante, it makes more sense now. :)

I was naturally suspicious of any supposed link between QM and consciousness, but purely as an explanation for increased computational speed it really is quite exciting! I suppose this might help explain how savants do such complex calculations so quickly as well as some of their other amazing mental feats.
cyghost (September 09, 2009, 15:52:47 PM):
Bravo, Mefiante. Solid explanations even I can follow and understand. Thank you for that.
Peter Grant (September 09, 2009, 23:07:10 PM):
OK, have been re-reading everything and have one, hopefully last, question. ;D

In what way would a quantum computer essentially differ from one of these:

http://en.wikipedia.org/wiki/Hybrid_computer

From what I've been reading they seem really cool, but unfortunately there seems to be little interest in them lately.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

Skeptic Forum Board Index

Non-mobile version of page