Thursday, December 18, 2008

The Emperor's New Mind

I recently re-read Roger Penrose's The Emperor's New Mind. I read it first in 1990, soon after it came out, and I still find it a stunning book, despite disagreeing with Penrose's speculations about consciousness somehow being boosted by quantum effect.

What makes the book so great in my view is that, in order to make his point, he has written an incredibly accessible textbook about a large number of topics in mathematics, physics and computer science. Penrose doesn't dumb down anything: he assumes the reader is an intelligent layperson, and explains each topic without compromise (though without getting into the technicalities of proofs).

Mathematical/computer topics include the theories of computability and algorithm complexity, Gödel's theorem, the Mandelbrot set, Turing machines (there's an appending with several pages of ones and zeros encoding a universal Turing machine -- I haven't checked it, but I'm sure that Penrose did), the Church-Turing thesis, and so on. In physics, he explains Maxwell's theory of electromagnetism (this was the first time I understood it), Phase space, special and general relativity, quantum theory, cosmology, black holes, space-time singularities and more. The table of contents reads like an overview of modern "hard" science. In the mean time Penrose also manages to explain Penrose tiling (the nonperiodic tiling of the plane, which he invented in the 1970s), the physics of nerve signals, Platonism, and important topics from the modern history of mathematics like Hilbert's program and Bertrand Russel's Principia Mathematica.

It's not surprising that Penrose shows prominently in the acknowledgments for Neil Stephenson's novel Anathem (in fact finding it there made me pick up my copy and start re-reading it).

Despite all this, I disagree with Penrose's claims about consciousness. Penrose does a good job separating his opinions on this topic from the science he presents, and in the end he cannot prove his hunches about consciousness any more than I can mine. Personally, I think that evolution is a perfectly fine explanation for our brains -- in other words, Richard Dawkins did a better job convincing me than Penrose. (I cannot recommend Dawkins' books enough -- you can skip The God Delusion though, just stick to his many excellent works about evolution and memes.)


ruiaf said...

"... despite disagreeing with Penrose's speculations about consciousness somehow being boosted by quantum effect."
I agree with you but one thing I never got is why isn't it a non-issue? Doesn't it just means that we can't copy (the current state of) a mind. Has it any implications whatsoever about being possible to do develop artificial consciousness? It's made of atoms anyway...

We should be thankful to Penrose, that's the science (divulgation) book!

ruiaf said...

Non-issue was an over statement. It's obviously important to know if we can do it in a touring machine. I just mean that it's more important to know if we can do it at all.

Dave said...

Why would you skip the God Delusion? I thought it a lucid, rational and well-argued explanation of a very common atheistic position (or "the position" if it's actually the case that there's only one.)

Calvin Spealman said...

I think this is going to be the first on my holiday reading list. It has been a long time since I've made time for recreational reading, and it is really nice to have such good titles to get back into it.

Guido van Rossum said...

@ruiaf: Penrose actually believes that the brain is a quantum computer, and that proper intelligence isn't possible without the performance boost that a quantum computer can get by simultaneously evaluating many alternatives. Personally I believe that while our brain is indeed very good at evaluating many alternatives, the reason is simply that it's built out of massively parallel hardware.

@Dave: Dawkins's other books always fill me with wonder and amazement about the beauty of nature in general and evolution in particular, teaching me how things work that I didn't know before. The God Delusion claims that religion is "just" a very successful meme, but I already knew that. Plus, it's particularly offensive for people with religious feelings (by design, but IMO unnecessarily so).

kkannike said...

You should read Penrose's "The Road to Reality" in which he first explains most of maths used in physics, beginning with numbers and ending with manifolds, tensor calculus etc. and then gives an overview of all of modern fundamental physics.

NA said...

I've never commented here, but I'm a big fan of Python. I want to suggest that you read Daniel Quinn's books: Ishmael, My Ishmael, The Story of B, and Beyond Civilization. They are not scientifically rigorous (for a good reason), but the ideas they present are influenced by Dawkins and are absolutely life-changing (at least for me). It presents a way of thinking that I find very clear-headed and practical.
He points out how the memes of our culture create its mythology and tell us who are are and how we came to be this way, and why we believe the things we do.
I hope you enjoy it!

Ram said...

Did you also read "Shadows of the mind"? Both the Emperors new mind and this book together present a really good case.

I feel the brain being equivalent to "quantum computer" or a massively parallel computer.. are essentially the same argument. [ walks like a duck quacks like a duck etc.. ]

jes said...

If our Selves were only a material machine that ceased to exist when the functioning parts required to make it work collapsed, than all human endeavor wouldn't be worth, for even if it was worth for the next generations, these generations would also only have a temporal existence. It would make sense only for our temporal success to work, and I think people in any situation of effort would be unmotivated contemplating this as a possible accepted truth. If this doctrine was disseminated, it would be very harmful for the human beings, for this would promote and explain selfishness. If all my existence is confined to here, wouldn't I do best if I get pleasure in every possible way? But we known this doesn't work that way, and so why civilization (mind-set) develops in the opposite direction of these beliefs?

The apparent incompatibility of evolution and religion certainly is a problem for people elucidation, IMO. I know that this is complex and not appropriate for debate in this place, but I think people should known that there is much much more evidence of consciousness than is covered. Please accept as my two cents to this topic a site called Spirit and Science (.org)


Guido van Rossum said...

Thanks for the recommendations -- I'll try to find time for them!

@Ram: I feel the brain being equivalent to "quantum computer" or a massively parallel computer.. are essentially the same argument. -- You should read up on quantum computers. They can simulate non-deterministic automata, which no "regular" parallel hardware can do.

@jes: It's a well-known fallacy that without religion humanity would degenerate into selfishness; Dawkins has some good things to say about this in God Delusion, and I see it every day -- many people I know are unreligious and they are not behaving any worse than the "faithful". (In fact the vote on Prop 8 indicates to me that it might well be the other way around.)

Finally, trying to prove God's existence (in whatever form) from "otherwise what would it be worth" is impossible -- who says there's an externally imposed goal to life or the universe?

(I expect you're utterly unconvinced by all this -- that's fine, the feeling is mutual. :-)

Ram said...

>> You should read up on quantum computers. They can simulate non-deterministic automata, which no "regular" parallel hardware can do.

There is no doubt about that!!
My reading comprehension a few years back and/memory could be very wrong but my understanding from Shadows of the mind was that, Sir Penrose postulates that it is not easy to separate a thinking sub component and acting sub component from an intelligent entity. [ Or may be I am trying to see what he says through a very colored lens] To that extent whether a thinking sub component of an intelligent entity is modeled as a deterministic system or not is moot.

I did not intend to say that massively parallel computers are equivalent to quantum computers.

dramenbejs said...

They can simulate non-deterministic automata, which no "regular" parallel hardware can do.

I must be missing something. All non-deterministic finite state automata COULD be provably simulated by nondeterministic f. state automata, e.g. by the most basic turing machine.

There is rigorose proof for that.

Where am I wrong? I would expect Penrose to be more careful in assertions than that.

dramenbejs said...

Oops! Turing machine is not finite state automaton, but the rest of what I wrote stands true.

Guido van Rossum said...

@dramenbejs: the difference is polynomial time vs. exponential time.

@Ram: I haven't read Penrose's newer works, but in Emperor he repeatedly makes the point that he finds it hard to believe that consciousness is "merely" an algorithm, and postulates that our brain uses quantum effects to beat the performance limitations of algorithms running on traditional hardware (parallel or not -- it makes no difference for complexity theory, while quantum computing would make a difference, IIUC).

My personal position is that (with e.g. Hofstadter, Dennet, and presumably Dawkins) I do believe that consciousness is "just another algorithm" running on "classic" (i.e. parallel but non-quantum) hardware. Evolution over hundreds of millions of years sounds quite a powerful engineering methodology if you ask me. I don't believe quantum computation (as it is currently being explored) is involved: even though quantum effects are clearly active of the scale of neurons and brain chemicals, I don't think it is involved in the actual computation that goes on in the brain.

Nick said...

Guido, if you haven't read any of Daniel Dennett's work on consciousness and possible ways it could have evolved, I think you'd like it (his book "Kinds of Minds" was an absolute revelation for me when I first read it years ago).

One of his more recent books ("Breaking the Spell") covers similar ground to that covered by Dawkins in "The God Delusion", but in a way which avoids the unnecessary antagonism that pervades the latter book.

@jes: the difficulty some folks may have with avoiding a spiral into nihilism without a belief in a transcendent aspect to the universe is the main reason I am far more sympathetic to spiritual belief systems than the perspective Dawkins conveys in The God Delusion. While I think it would be better if everyone could be appreciative of the beauty of the universe and believe that life is worth living (and living well) without needing an external factor such as 'God' or 'the spirits of our ancestors' to justify it, the human animal just isn't built in a way that makes that easy.

nateø said...

I'm a bit late on this post--found your blog while learning about Python (which I really enjoy; nice work Guido!). To Nick's point, I think that the 'spiral to nihilism' is a logical implication of naturalism/materialsm. This is seen in modern Anglo/American philosophy, with the progression of Quine to Richard Rorty, who has the intellectual honesty to say that he cannot give reason for the way he things things are/should be... In short, I am unconvinced that analytic philosophy/science alone can deliver certitude. This, I think, has the force of a reductio, and gives us good reason to look back to the classical tradition (e.g. Aristotle, Aquinas, et. al.) for a philosophical framework that can deliver certitude. (I've heard that Alisdair Macintyre makes a good case for this in Whose Justice? Which Rationality?)

giannisfs said...

As a general rule i believe we all assume that once someone uses cloud computing without prior experience on setting up his own server his has become a consumer , a client or simply someone depended.This is not a problem by it self , it only happens to be problematic when the big company (cloud) is coming to an end or when it start acting arrogant and asks for fees.Show probably the healthiest solution to which (cloud or home) computing one shall prefer is at least to get to know how can he run his own home server.
If you are a company or just a person it will be always a great advantage having at least the ability to set up you own server , your little exit of danger.Also people need jobs , the more cloud computing dominates the less PC-people will find job easy(but not exactly of course ). So let's get hands on setting up our own server ..just in case..
Thumbs up for python ;)

Tim said...

God's lack of existence is a very successful but restrictive meme. If you believe "God IS the sunset", as well as "God CREATED the sunset", the statement "There is no god" is similar to "There is no reality".

Furthermore, all science (and I consider myself a scientist) is based on faith. Eg. Pick a number of dimensions. 3? 4? 11? Now PROVE that that is the definitive amount of dimensions. You cannot do it. Meanwhile, all of physics is built on a faith in a fixed number of dimensions. F=M*A has underlying assumptions regarding numbers of dimensions.

Dawkins is tired, but I will re-read Penrose. Have you read Seth LLoyd's 'Programming the Universe: A quantum computer scientist takes on the cosmos'?

valjok said...

I completely agree with your appraisal. Among the educating coverage of paradoxes, the idea of quantum origins of mind seems to be pursued with the help of great omission: explained why the symbolic (formal determinstic) logic is inconsistent for modeling our mind, he ignores the true source of computational power - the product of huge NEURAL network CONNECTIONISM known as THE INTUITION, which is analogue at that! Trying to use individual neurons as binary gates to build a logic circuit is ridiculous for anyone who is familiar of NN basics!

Yet, the magic of intuition is analogous to quantum effects since 1) all incompletely certain alternative possibilities are pondered and the answer is given immediately is similar to ensemble crystallization due to entanglement and 2) the computation by the myriads of simultaneous inter-neuron communications (randomly affected by thermal noise at that) is almost impossible to follow. Nevertheless, the neural diffusion of computation into the uttermost form of parallelism takes place at the classical level - the "quantum noise" is not necessary for miracles of thought! Secondly, NNs are adjustable to produce reasonable answers. Why to entail the quantum indeterminism?

Guido van Rossum said...

I feel compelled to note that while I am moderating comments, I will allow any comments to be posted that aren't spam, regardless of whether I agree or not.

Also, someone recommended Seth Lloyd's 'Programming the Universe. I have since read it and it is fascinating but it falls short as a science book. (No index? Really??) I'm still skeptical of quantum computing as the road to more computer power in the conventional sense, but clearly quantum theory is some kind of key to understanding the universe, and it we're not getting it yet. Hopefully more about this at some point (after I finish Penrose's The Road To Reality maybe).

Tim said...

Seth's book is kinda pop-science but that is about the amount of cycles I can spare. I think QC will be here fairly quickly (and offer no insights into consciousness). But who knows. You were right about map-reduce.

If you remain interested in the moment of wave function collapse, check out Julian Barbour. He takes it too far but his basic premise is insightful. And thanks for reminding us that Penrose is still relevant.

valjok said...

The blog host's reaction leaves impression that I am a lickspittle, despite I explained the important thing - why exactly we have the full right to be skeptical on don Penrose's line.

Let me elaborate. He tells that the mind needs quantum entanglement of large ensembles to realize the ideas in entirety (aka quasicrystals are formed). Nevertheless, I started my study of NNs by the simplest Hopfield classifier. You can take some short binary vector and see how quickly, in couple of steps, it converges to a pattern that is predefined by the network weights. The neuroscience considers the output production as "energy minimization". So, no quantum effects are necessary to minimize energy at large scales. Yet, Penrose prefers not to speak about neural models.

Might be he implies that NNs are still classical models, necessarily built on top of strong AI. I would prefer more elaboration on this matter than Minsky's mocking "brains are computers made of meat".