Tuesday, August 29, 2006

The human use of quantum human beings

I've been back in the US for a while, in that part of the country
which is the real thing, made of towns where people can only move
around on wheels, where elsewhere respectable bookstore chains sell
mostly bibles, and where the countryside creeps in all the way to
the scattered shopping malls that constitute the decentralized
focus of the town experience. No downtown, no highrises, no subway
system, none of the familiar city experience. It is all so far away
from the years when I lived in this same country in the real cities,
when I would sit at night on a terrace high above the cityscape,
feeling its pulse, its voice through the cold night air: the harbor,
the construction work, the city lights. This now is the other America,
the real one. The weather here is good for big reptiles: a hot and
humid swamp (ever read Ballard's "the drowned world"?). I surely don't
mind the tropical heat, and I resent the air conditioning much more
that the native weather. It's hurricane season, but this year it's
been unusually quiet so far. Ernesto is lazily making his way up the Gulf,
hesitating between hurricane and tropical storm.

I am trying to put together a proposal for an NSF grant, trying
to make my improbable mathematical musings sound like something
a government might want to spend money on. Hopeless.

I acquire a different personality in different places. My taste
for things changes. Around here I like to wander in small second
hand bookstores and try to fish out the unusual. I become strangely
attracted to "vintage sci-fi" and I find special pride in discovering
rare copies of the Soviet or British writers. This time I managed
to get two Strugatsky novels, one Kirill Bulychev collection and
two Fred Hoyle. Ah and the Criterion Collection put out a new DVD
edition of "the man who fell on earth". That's not bad either.
I like electronics here too, sold everywhere as readily available
cheap mass products. Why do I keep feeling at home in this country,
despite of everything? Perhaps it is because this is a place that
does not require people to have a well defined identity: the melting
pot that other parts of the world are utterly unable to create.

Browsing randomly (or godly, see below) among the shelves of the local
Boreders Bookstore (of the type that, in this part of the country, leans
heavily towards the "inspirational" side), I happened upon a title that
momentarily caught my eye: "The quantum brain" by Jeffrey Satinover.

The title immediately reminded me of the worst kind of new age "quantum speculations". Having expected the worst, I was positively surprised when, opening a random page, I landed upon a very honest and well delivered explanation of the 3-neuron Hopfield neural network. I couldn't help thinking that the general aspect of the book looked suspicious, but if a book is able to argue an unsustainable point of view with sufficiently well argumented and intriguing cleverness, it might be worth the ride. After all, it seemed that, for a good part, it consisted of a decent exposition of interesting facts of neuroscience, mathematics, and quantum physics, all of which rank quite highly among my favorite readings. I do not normally read popular science, for the reason that it puts way to much emphasis on extraordinary claims and far too little on the extraordinary evidence. In this case, however, the unusual appearance of the book convinced me to make an exception.

The first half of the book makes for a very pleasant reading. It is full of interesting information about the early days of neural network theory: the Perceptron trying to model the retina, Martin Gardner's Hexapawn, the debate on the XOR logical gate and its realizability by a multilayered Perceptron, NetTalk and the birth of speech synthetizers, convergence of networks and the effect of "pruning" the connections, canonical microcircuits of the cerebral cortex (Kohonen's self organizing maps), spin glasses, Hopfield nets, basins of attractions, energy landscapes, annealing techniques to achieve minimal energy configurations, cellular automata (Life and the like), Belousov-Zhabotinsky oscillating chemical reactions, genetic algorithms, evolutionary hardware. A veritable tour de force across many of the buzz words of neural and cognitive science research. Main conclusions: neural networks provide systems that closely resemble the functioning of the brain in a biologically realistic way. Main question asked: are we then just computing machines? I do not truly get why a straight "yes" answer to such a question should be so upsetting to some, nor why one necessarily needs the "just" in the question. Still, this whole part is nicely presented: a bit too much on the "popularization" style perhaps for my taste, but with enough attention paid as to not avoid misrepresenting the original scientific sources.

Now comes then the second half of the book, where the author presents his case for a "no" answer to the previous question. Given that all seems to point to an easy hand for the "yes" camp (which the author is well aware of), it will require considerable cleverness and skills in presenting anything close to a convincing argument in favor of the cause the author chooses to advocate. The author approaches the question at first through a seemingly innocent observation, near the end of the first part of the book, where he reports on two research papers exhibiting systems where quantum tunneling greatly enhances the annealing process and speeds up the convergence to the minimum energy configuration. That's the starting point for Part 2, which is aptly called "Miracles". The choice of title is reminiscent of the old Far Side cartoon showing the mathematician standing at the board, trying to present the proof of a theorem, where in the middle of the argument in between formulas one finds in big writing the crucial step of the argument "and then a miracle occurs". Satinover's argument in the second half of the book is very much of this nature. He would like to argue that quantum phenomena happening in the brain get amplified to the macroscopic level and are responsible for the presence of an element of randomness which frees the human brain from its mechanistic destiny and allows for the possibility of free will (and for other possibilities I will return to discuss momentarily). Others have walked down this venue before, including Penrose's own attempt at proposing the existence of amplified quantum effects in the brain, not to mention the good Bohm of old. To be true to the author, I should say that, in my opinion, although I do not believe for a moment in the conclusion he draws, his case is argued much better than the Penrose attempt.

The beginning of Part 2 is spent on the usual story of the "paradoxes" of quantum mechanics popular books are so fond of. Here I feel like making a general comment of sorts. Should we really keep propagating a view of things which is about a century old? Quantum mechanics is a wonderful subject, no question, but I can hardly think of a working physicists today who has not come to terms with it and does not feel comfortable about it. Propagating the idea that physicists today lose their sleep over the philosophical meaning of the two slit experiment simply does not do a good service to the community. It is much like the case of those popular books that portrait non-euclidean geometries as being a shocking revelation to the world of mathematics. They surely might have been at some point in history, but at present mathematics has moved so far beyond that stage that one can hardly conceive that there was ever a problem at all. Pretending otherwise would be like portraying the modern sky watcher as shocked by the idea of the earth revolving around the sun.
Quantum theory evolved far beyond quantum mechanics, with highlights like quantum field theory, gauge theories, and the standard model of elementary particle physics to its credit, which do indeed have far more intriguing and mysterious aspects to them than the poor Schroedinger feline. This is of course just a general observation aimed at the whole literary genre of quantum fiction and not just at the present book. This book in itself does actually a pretty straight and honest job at presenting the usual mantra of the mysterious nature of quantum physics with good clarity and without getting carried away too much. In fact, the example of the ammonia molecule to illustrate tunneling and equilibria of a quantum mechanical system is well chosen.

The subsequent part on information and quantum computing is interesting, though perhaps something more precise could have been said on the topic of quantum computing ("Zen and the art of quantum computation" admittedly makes for a nice title). There follows a discussion of Penrose's viewpoint (that of "Shadows of the mind") on quantum effects in the brain. The criticism is well taken. There starts then an interesting discussion on microtubules structures. This is in my opinion the nicest story you'll find in the second part of the book. First looking at microtubules structures in the cilia of the paramecium, then at the modeling of microtubules as networks with computational power, then on to protein folding ("organic origami") and the problem of long-distance electron and hydrogen tunneling in organic molecules. This section builds up most of the ingredients the author will need to "argue" his case for quantum effects, all he needs next is to describe a plausible mechanism for amplification.

This he finds in another favorite of science popularization (after the Schroedinger cat), namely chaos theory. The author is careful in distinguishing the appearent randomness of classical chaos (an entirely deterministic phenomenon) from the true randomness of quantum mechanics. The brief overview of classical chaos is well delivered, with butterflies kept to a minimum and more sincere mathematical examples exhibited. Enters quantum chaos, a more subtle type of phenomenon, which he author only discusses briefly (another point where more details would have been welcome). The author argues, based on the example of coupled quantum oscillators and of billiards, that quantum chaos can reinforce Poicare' recurrence instead of disrupting it and can enhance convergence of the system to a stable attractor. There, the auhor argues, is a possible amplification mechanism from the quantum to the macroscopic scale. Whether this is really what it is is quite unclear to me, but so far at least there has been a lot of intelligent discourse about the science side of things.

So far so good, but then a miracle occurs and a huge leap to some kind of conclusion is taken. I personally recommend to skip entirely the final Chapter 17, which as a clear warning to the reader starts out with a section entitled "god". This is where the book goes down the mystical drain. If one wishes to replace in the vocabulary the word "randomness" with a new word one calls "god", that's quite all right, as long as this term is not loaded with meaning that was not contained in the original term. In other words, stochastics by any other name would be just as aleatory. Then why not save a step and just remain with "chance", which at least is a well defined mathematical concept and not a murky metaphor. The mathematical proof for the impossibility of hidden variables in quantum mechanics appears to rule out the possibility that the new term be something other than a simple renaming of randomness, in the sense proposed by the author in the beginning of his final chapter. I do not want to be too harsh with the author, after all I read the book with pleasure and I wish that more of the similarly inclined people would argue their point with such class and cleverness.

In summary, would I recommend the book? Well, I certainly would: the "religious" overtones might upset you, but the book is well written and it does make for an entertaining bedtime reading. It is a good fairytale of the 21st century.

For those who would like to read something that can really be considered profound on issues such as determinist and free will, physical and biological systems, written by a leading scientist with a broad humanistic perspective, please read Norbert Wiener. His "The human use of human beings", also known as "cybernetics and society" was written in 1954, but despite the dramatic scientific and technological progress, especially in fields such as neural and cognitive science, and the many changes that both cybernetics and society underwent since that time, it hasn't aged a bit.