Comment by bloak | Hacker News Reader

Comment by bloak | Hacker News Reader

"Conway thinks the free decisions of the particles inside us might account for our own free will."

Another mathematician, Roger Penrose, made a similar claim, but with "consciousness" instead of "free will".

I find this very implausible. So does Douglas Hofstadter, probably, though I've not yet read his main book on the subject.

Thought experiment: You make a mathematical model of how neurons work, you cut up a human's brain, and you build a machine that simulates it. Wherever the model required randomness, you use pseudorandom numbers, generated cryptographically, so the entire system is deterministic. You connect it up to some kind of humanoid robot. Does the resulting system behave roughly like a human, and, if so, does it have free will, and is it conscious?

People who believe that consciousness or free will depends on fundamental physics have to believe:

* Either that the simulated human won't work like a human: it will mysteriously fail to function, but how exactly?

* Or that the simulated human will seem to work like a real human, but nevertheless it won't be conscious; it will be a mere "zombie".

The first option seems intuitively implausible to me, and the second option sounds like silly sophistry.

> I find this very implausible. So does Douglas Hofstadter

Dennett does a good refutation of Penrose. There's some in Darwin's Dangerous Idea (I think; might be confusing it with Consciousness Explained), but he finishes the job in Freedom Evolves

Long story short, his assertion is that the contradiction between determinism in physics and free will in humans is a category mix-up, not a true contradiction. There are so many layers between the two levels, so many interactions, that when you get to the top level at which we consider an coherent blob of "matter" called an "organism" -- with a "brain" even -- there really are things called "thinking" and "free will". But way down at the level of molecules, those concepts aren't even nonsensical; they're just meaningless. To talk about free will, you have to have a very high vantage point. And that's why it gets confusing when you go back down to the building blocks.

This is similar to the systems argument against Searle's Chinese Room: the phenomenon we find so interesting disappears when the system is broken down, exactly because it's an emergent product of the system.

There are analogies I could make, but I'd probably miss something, so I'll leave it to Dennett (he actually uses the Game of Life to make his point). I highly recommend reading Freedom Evolves if you're into these ideas. Very thought-provoking, especially for its compact length.

In the case of free will I don't see any intuitive implausibility. It's intuitively plausible that someone/something could exhibit human-like behavior without actually having free will. After all, some people think that regular humans fall into this category!

The issue is that a lot of people look at humans and say "This kind of behavior requires and proves free will", which would mean that exhibiting human-like behavior without free will ought to be impossible.

I don't think anyone concludes that humans have free will on the basis of their behavior. At least, that's a pretty unusual line of argument.

Maybe not with the wording I used, but I'd say it's actually THE most common argument around. The whole "machines can't love" line.

Most people who argue for free will do so on the basis that it appears that we are free, not because there is some physical mechanic of the universe that implies free will.

The difficultly in deciding whether humans have "free will" suggests that the concept of "free will" might be silly sophistry.

Similarly, the difficulty in deciding whether or not Goldbach's conjecture holds shows that the concept of "prime number" might be silly sophistry.

Hardly worth replying, but I don't see the similarity. A 6-year-old understands perfectly well what a prime number is, and you can write a program to test for primality in half a line of Haskell (though the performance will not be good). Now what's a "free will"?

I was responding to your original claim, which appeared to rest on the assumption that problems which are difficult to solve are likely to be psuedoproblems. I deliberately chose a difficult problem based on a concept that has a clear definition (prime number) in order to show how silly this line of argument is.

If you are just saying that it's difficult to come up with a generally agreed upon definition of "free will", then that's of course true. But that doesn't mean that there aren't any genuine philosophical problems relating to free will. By way of analogy, it's also difficult to define terms like "property", "freedom" and "constitution", but that doesn't mean that all of political philosophy is sophistry.

They probably would be conscious too if QM randomness is psuedorandomness. It is not known whether Quantum Randomness is special. It could be the result of deterministic pseudorandomness. All randomness might be either complex scenarios with insufficient information or algorithmically random sequences..

Here is a good in depth article by Scott Aaronson on this: https://www.americanscientist.org/article/quantum-randomness

> if QM randomness is psuedorandomness

Isn't that the same as the hidden variables theory which has been proved wrong experimentally?

Local hidden variables have been disproven. Non-local ones are an entirely valid solution.

a) Penrose' claim is about something completely different, namely the non-algorithmic computational abilities of the brain by quantum-gravity effects

b) intuition is irrelevant - this is a proof of the physics of free-will, it doesn't matter how 'plausible' it sounds. Quantum mechanics isn't intuitive either - why should it be ?

a) Penrose has probably made a lot of claims over the years. I may be thinking of earlier writings than you are.

b) Proofs tend to be a garbage-in-garbage-out sort of thing. It's best to check the output against your intuition because you may have made a mistake in your reasoning or, more likely, in your assumptions and definitions. Intuition is not fixed and immutable: you can develop it to include things like quantum mechanics, probably.

It seems to me your assertions assumes that the randomness you get out of a quantum mechanical system is essentially same as psuedo-random sequences in quality. which may very well be true functionally (as in being able to simulate specific behaviors) but it is still completely deterministic (unless you use a true random event like decay of a nuclei etc). Until then your option-2 fall into the phys-zombie mode.

If I peer deeper into that line of thought I'd say it assumes that consciousness is simulate-able by physical deterministic processes and then goes onto say Ah look the behavior is indistinguishable so we have simulated the consciousness. BTW you can tell I am not a believer in turing tests like setup for 'measuring consciousness'. IMO its of a completely different nature.

here is my take: Could it be that consciousness needs quantum mechanics to actually 'express' itself in human brain? this is borrowed from Penrose's Orchestrated-Objective-Reduction thesis. if so we could have either of those outcomes depending on the sophistication of your simulation and handling of various perturbations in human brain architecture. personally I lean more towards this option, especially after learning that QM processes play more central role in biology than we thought originally despite 'it' being too warm, wet & noisy. In short, if there is any evolutionary advantage to be had from QM effects, biology will adapt to exploit it.

I'd argue Option 2 is not silly sophistry if you consider following scenario:

1) Lets say we create a planet sized computer with enough capacity to simulate every human being in the way you described. then we decide Ah we dont need to be in meatspace anymore because what difference it makes anyways.. so we simulate all human minds and 'upload' each human into this computer. now if we are wrong about consciousness then essentially we would have willfully committed genocide on our own race (& every other living thing).

2) suppose we develop a Opt-2 based super-intelligent AI which is way better then us at everything. now why should it keep such resource hogs as humans around? IMO precise definition of consciousness matters quite a bit here as it creates a quality worth preserving/exploring further.

Reply on news.ycombinator.com