on February 29, 2004
To all those who wish to dismiss this book: Let's give Roger Penrose a break. After all, he's pretty smart (ahem!), and even if he turns out to be incorrect in suggesting that consciousness can be explained physically using physics we don't have yet, the book is a vigorous and entertaining attempt to put forth the case. He states up front that _we don't have the physics yet_, so where's the controversy?
I find the claim that Penrose simply rejects the view that the mind is a (computational) system, because no system can be both consistent and complete, a little misleading and certainly no substitute for reading the book. To address this on just one front, there is also a positive side to Penrose's argument, namely, that the mathematical insight needed to recognize undecidability and related arguments as legitimate--an insight he tries to defend against competing philosophies of math--would itself appear to lie outside the realm of computation.
As for the idea that ENM is a poor man's GEB, I see the two books as completely different in motivation. In GEB, Goedel is central in leading to the conclusion that some sort of self-reference lies at the heart of intelligence. In ENM, Goedel is important in flushing out regions of mathematical thought that appear to be non-computational, but the overarching suggestion is that consciousness will someday be explained using as-yet-undiscovered physics.
For me, the attractiveness of both books lies in their "vigor with rigor," that combination of mastery, humility, and generosity one longs for in science writing.
on December 18, 2003
Roger Penrose, "one of the world's most knowledgeable and creative mathematical physicists," presents in his 1989 Emperor's New Mind one of the most intriguing and substantive popularizations of mathematical logic and physical theory that has ever been published. As a reader of many books written by scientists, I will say that few compare with this one. Penrose wrestles with what he sees as some of science's most inadequate or poorly developed (although popularly accepted) ideas. As certain physical theories are found wanting, his grapplings extend to some of the deepest questions of metaphysics. Of the deepest questions, Penrose says, "To ask for definitive answers to such grandiose questions would, of course, be a tall order. Such answers I cannot provide; nor can anyone else, though some may try to impress us with their guesses." While he speaks respectfully of individuals with whom he has certain differences of opinion, the "some" in that statement might be taken to be Hawking, Dawkins, Dennett, to suggest a few. The author here tends toward a more humble and questioning approach. Penrose's puzzlings are complex and his positions are sometimes misrepresented by even his admirers. A case in point may be the fact that he finds cosmic inflation theories to have less explanatory power than others claim for them -- this doesn't mean he necessarily rejects inflation, rather he doubts claims that inflation actually helps explain the specialness of the early universe. Positivists may be disposed to discount the problem but there appears to be good reason for Penrose's skepticism. However this is not treated in this volume.
Rigorously building a case against the fundamental arguments for strong AI, Penrose begins with what for him is to ultimately be 'le coup de grâce', considerations and arguments from mathematical logic. If the human mind works non-algorithmically, then we know of no way to digitize/program its processes. The mind does in fact function non-algorithmically, a fact demonstrated without much difficulty. It learns in intuitive, non-linear, and mysteriously creative ways. The idea that some non-algorithmic approach might achieve a program equivalent to the human mind is not supported by any "useful" (or better) physical theory and is not mathematically tenable. Strong AI is thus relegated to a mere ideological preference and to sci-fi. In his mathematical considerations, Penrose is most interested in the work of Turing and Gödel and in the Platonic essence of mathematics itself. Concluding that the human mind cannot be reduced to an algorithm (or any set of algorithms), Penrose next questions whether the mind might be reducible physically. Here he finds the questions and answers less well defined than he has in mathematics. His tour of classical and quantum physics features interpretations and ideas that many readers may have not encountered (which makes the text fun). The problem of "correct quantum gravity" (that is, the incompleteness [or incorrectness?] of relativity and quantum theories) is one that Penrose and other theoreticians have struggled with for decades. Penrose wonders if this mysterious and conspicuously missing physical theory might be related to the also conspicuously missing science of mind ("the mind-body problem"). This speculation on his part is the theme also of his more recent books. As Erwin Schrödinger (like Einstein and Gödel, Platonists all) seems to be one whose ideas are of particular interest to Penrose, I will cite Schrödinger's view: "Consciousness cannot be accounted for in physical terms. For consciousness is absolutely fundamental. It cannot be accounted for in terms of anything else." But Penrose doesn't quite argue this view, although it would seem an obvious conclusion from his best arguments! Here is a classic example of how we may know something without knowing everything: we can know that the human mind cannot be reduced to an algorithm -- or algorithm of algorithms -- and yet it is not known that we can even know what exactly mind is. Particularly so if, as Schrödinger says, mind is irreducible.
The chapter on cosmology is excellent, as one might expect of a Roger Penrose. The consideration of the "specialness" of the initial [cosmological] conditions and of the relationship of this specialness to the second law of thermodynamics is also fascinating as it is precisely the second law that lends the "arrow of time" its apparent non-symmetrical aspect -- in other words, defines physical "reality" as we experience it. In this sense, the second law connects the human mind to the cosmos (which is interesting but does nothing to help us "reduce" mind).
Penrose suggests, and I cannot find any reason to disagree, that all scientific theories can be assigned to one of three broad categories, which he calls: (1.) SUPERB, (2.) USEFUL, (3.) TENTATIVE. All SUPERB theories (there are roughly a dozen) stand within the purvey of physics, and: "It is remarkable that all the SUPERB theories of Nature have proved to be extraordinarily fertile as sources of mathematical ideas. There is a deep and beautiful mystery in this fact: that these superbly accurate theories are also extraordinarily fruitful simply as mathematics. No doubt this is telling us something profound about the connections between the real world of our physical experiences and the Platonic world of mathematics." Over time, theories (particularly those that do not feature such mathematical beauty) may tend to move between the categories. Theories held to be SUPERB for centuries have dropped completely from the current categories, theories have faded and re-emerged. . . "we should not be too complacent that the pictures that we have formed at any one time are not to be overturned by some later and deeper view."
Some readers will not like the fact that, after extensive rumination on very difficult and deep questions (like "what is mind?"), the author doesn't conclude with a pretense that he, or anyone else, has definitive answers. This reader appreciated the integrity of Penrose's questionings and of his conclusions (or lack of conclusions). I will misappropriate one of Penrose's terms -- as a text examining mathematics, physics, and the human mind, this volume is SUPERB.
on September 24, 2003
I wish only to comment on the context of the book, and only because it has become such an important book. Really what I want to say here is: there is no way to pigeon hole this author or the book. This isn't simply a controversial book, this man asks the kinds of questions that change things, and the questions he has asked here have turned the claims of his opponents into so much philogestion and aether. He is starting from a point where so many others were prepared to do nothing but offer gratuitous assumptions and when challenged have only responded dogmatically. Be on guard for prior claims, especially from the life sciences.
Some have said Penrose is a self professed materialists. But unless we are talking about the Marxian kind I can't see that. Whatever his viewpoint it does not seem overly compatible with billiard balls and atomism. Penrose might as well be called an immaterialist, an idealist etc... I don't think his ideological opponents cannot simply claim him and expect the issues to go away.
This book doesn't seem overly compatible with gene-neuron-evolution-drug-patent-status quo rhetoric. There is probably no easy way to translate Penrose into Spencer. It is possible that many of the people who want to dismiss Penrose are people who would have wasted the rest of their professional lives and vast amounts of public monies on superficial or rather convenient dead ends. Penrose may not be the only one doing good work but I am guessing that many of his critics are not going to be able to continue on with business as usual. Their capacity to judge is being called into question, but more than that their collective character has been reflected in their take on the world- somehow Penrose has made that obvious. Many look pretty superficial and credit oriented. They are loosing credibility- and history is probably taking note.
Questions need to be asked. We have been fed false dichotomies, excuses (chance), empty genomes etc... Yet none of the basic questions involving, for instance, 'life' have been honestly approached. What is Life- even as a category or non-category? How does development, morphology, tissue regeneration come about? All we get is biotech marketing hype; which has come to pass for education in some countries. What about cancer and aging(?)- nothing, certainly nothing honest. And lastly what about consciousness? The people trying to denigrate Penrose with the 'Platonism' label are probably coming from the viewpoint of scientism, or at least the economically inspired laziness of the establishment.
on March 23, 2000
First, what this book is not: It is not "creation science"...it doesn't address evolution...or the existence of God...or existence of the human soul. In other words, it is NOT special pleading against modern science by someone with a religious agenda. What it IS rather, is a solid study of cognition, theories of artificial intelligence, and the enduring problem of the nature of human consciousness by one of the world's top physicists (a professed materialist by the way, not a religious believer), who together with Stephen Hawking developed the astrophysics of "black holes" in the '60's. What Penrose suggests here (a theory he expands on in his subsequent "Shadows of the Mind"), is that science, and specifically physics, is inadequate now, and more importantly will always be inadequate, to describe the nature of human intelligence, cognition, and consciousness--a thesis similar to the showing of Godel's 1931 Theorem that certain fundamental axioms of mathematics were incapable of proof within any mathematical system. In other words, Penrose suggests that there are elemental restrictions within science itself limiting our understanding of our own mental processes, which concomitantly limit the possibilities for development of artificial intelligence. And that obviously doesn't sit well with those for whom naturalistic science is itself a kind of "religion," as some of the dismissive reviews on this page show. My advice: just ignore them and read this book, and well as its successor, "Shadows of the Mind." It's a challenging read and not for intellectual lightweights, but it will richly reward those with the patience to make it through.
on January 14, 1997
This book, together with its companion Shadows of the Mind, is the
result of many years thought by one of Britain's most original scientific
thinkers. It contains a fantastic sweep through classical and quantum
physics as well as Godel's theorem, Turing machines, and the like.
His conclusion - the mind is not governed purely by algorithmic processes -
is highly unpopular with many philosophers and the AI community. However it
is very carefully argued and apart from anything else makes a significant
contribution by laying out in a very clear way the logical options
available in understanding aspects of how the mind might compute.
Many of those working on understanding the mind do not want to be told
they will have to get deeply involved in quantum mechanical issues
before they will get anywhere near their goal. However Penrose makes
a profound argument that this is in fact the case
on October 28, 2003
Penrose is going after all those who thing that a computer will someday be able to "think" like we do. In other words, he does not believe that - because of some rather esoteric quantum effects - that our electr-organic brain can be replicated by electro-silicate.
This does not mean that computers will not be able to mimic, to respond, to act in a way that one has no idea if the person they are talking to is a machine or human. All this is possible, even probable. But Penrose is on a mission to raise human consciousness above machine performance - or rather, to demonstrate that it is of a different kind rather than a different order.
My only problem with his analysis is that we simply cannot know what may or may not happen in the future as technologies merge and grow and intertwine. With current technology there is not a chance that a PC will some day "recognize itself". But that is not the question really; everyone knows this. The real question is what does it mean to be human, what is consciousness, and can these characteristics, traits and components be reborn outside of organic matter?
Interesting, sometimes difficult read.
on October 18, 2003
I have a simple classification for authors of science books. A few of them are giants. Several respectable scientists are professionals. Many authors of science books are dwarves. Giants produce books that are great, useful or useless. Professionals produce books that are useful or useless. Dwarves only produce useless books. For instance, D. R. Hofstadter is a dwarf: all of his books are pompous and useless (Pulitzer Prize winning or not). So are Michio Kaku's books: useless all. I can't believe that anybody could compare these two authors to Roger Penrose.
Penrose is a giant! He is one of the world's greatest mathematicians and a pretty good mathematical physicist and philosopher. Still the author's greatness does not assure that the book is itself great. In fact I used to believe that this book was useless. I believed this until I actually read it! It is at least very useful (5 stars), even to professional scientists and mathematicians. It may be too difficult, in places, for an average reader. But, by simply skipping the hard parts, even the average reader should get a lot out of it.
Let me explain what is so great about this book. It is a voyage to the borderlands of physics, logic and mathematics, in search of cracks in the naïve materialism that seems to be (again!) the orthodox philosophy (or religion?) of the scientifically "educated". Very many pearls are collected during this voyage. One just has to look at Figure 4.3 (hardback edition), showing the "space" of propositions in a formal system. The fractal geometry of the set of provable propositions (which is not recursive) is shown. Also shown is the even stranger set of the true propositions, which is not even recursively enumerable. I can't avoid thinking that this figure illustrates the logic of the future: much less formal and essentially geometric in nature.
The discussion of Quantum Mechanics is also very good, because it explains (at the beginning of Chapter 6) the extreme relevance of QM to everyday life. I will quote from Penrose: "...The very existence of solid bodies, the strengths and physical properties of materials, the nature of chemistry, [...] the reliability of inheritance - these, and many other familiar properties, require the quantum theory for their explanation. [...] we, as sentient beings MUST live in a quantum world..." Anybody for whom these facts are not obvious is simply scientifically illiterate. Quantum theory is not an exoteric theory about rare phenomena, but an explanation of why the world we live in contains things, solid objects, like chairs, viruses and cats. Every computer is a quantum computer, in some sense, because classical physics cannot even explain the existence of solid objects. So even supporters of strong AI have to admit that, in a weak sense, consciousness has something to do with quantum mechanics.
Penrose's is one of the few popular science books that stresses the fact that quantum mechanics makes predictable machines possible, that they cannot exist in a purely classical world. Most popular books stress the indeterministic nature of QM, rather than its stabilizing effects. I will repeat it, again, in Penrose's words (from Chapter 5): "...'Solidity' for an object itself composed of many particles is something that really needs quantum mechanics in order to work. It seems that even a 'classical' computing machine must borrow from the effects of quantum physics if it is to function effectively!"
Because of its importance, the conceptual problems of QM are the main scientific and philosophical problems of our time (and, maybe, of all times). Penrose thinks that the solution of these problems lies in the fact that "ordinary" quantum theory breaks down when confronted by gravity. In particular quantum theory stops working when it would produce a superposition of different geometries in a quantum theory of general relativity. This solution of the problems of quantum theory is at least economical, because quantum gravity needs work anyway! Penrose thinks that CQG (Correct Quantum Gravity) will solve the obvious problems of quantum theory and also provide insight in the problem of consciousness. This is very speculative, but he could be right. He is certainly more likely to be right than any of the "great thinkers" that actually deny that there are conceptual problems in quantum theory!
Penrose is, of course, wrong about several important subjects. For one he is (was?) an opponent of inflationary cosmology. This theory has now been confirmed to a very high level of accuracy by cosmic-background-radiation experiments. Also Penrose's detailed theories of brain function (exposed in successive works) are obviously wrong. He should have known better than proposing detailed physiological models, given his obvious ignorance of biology.
However all of this does not matter, because consciousness is too important a problem to be left to brain scientists or even to biologists. It is still a philosophical problem. Then the question is: how successful is the book in its main aim? Does it disprove the strong AI assumption? I think very much so, since strong AI is an extraordinary assertion that lacks even weak evidence, much less the extraordinary evidence it actually needs. This book is not a (unnecessary) refutation of strong AI, but an attempt to find a replacement, while avoiding total irrationality.
on April 16, 2002
As other reviews have mentioned, Penrose attempts to demonstrate the impossibility of strong AI's thesis. In the end, Penrose presents two core arguments: (1) a well-reasoned, deductive, and logical proof based on Godel's incompleteness theorem, and (2) a complete conjecture about quantum gravity that has little to no supporting evidence.
Thus, please save yourself the time and just read chapters 1-4 and 9-10. The 225 pages that are chapters 5-8 give the background for Penrose's second argument. These pages are mostly filled with erudite, technical physics. The first four chapters, on the other hand, give an enlightening review of artificial intelligence and outline the mathematics needed to follow his first core argument.
Personally, I find both arguments unconvincing. Not being a mathematician, I won't embarass myself by trying to show why Godel's incompleteness thereom is not entirely appropriate for a discussion on artificial intelligence. As for the second argument, neither you nor I need worry; it is entirely useless.
on March 15, 2002
I will be quick in this review.
First, the physics are stimulating, the brain science lacking, the speculation interesting. But the aruments are absolutely and completely flawed. Its not just that quantum effects almost certianly have no effects on consicousness, or cognition for that matter, and the argument implicitly made: "quantum is mysterious, consciousness is mysterious, so they are interrelated" is ridiculous.
It all starts with the church-turing thesis. Any algorithmically computabele process, can be carried out by an universal turing macine (for our purposes, a computer). Now the idea is too figure out if computers can have a mind like a brain has one. Penrose holds that it is not posible, so he asks the right question: "is the human mind algorithmically computable?". Penrose says "no", and his reasons are simple: humans can see the truth of godel propositions, and human mathematicians have sudden "insights" that are, well, supposedly non-computable. The first thing one can do, is, well, hold that in fact the human mind is indeed computable. The truth is that this is a pretty fair bet. Just look at the neural-network progress made in PDP. Actually it is on Penrose to prove us that the essentials of mind and consciousness (not godel propositions)are non-computable. But at the end, Penrose seems to beg the question.
Now the Godel argument is a little bit more straightforward, but wrong nontheless. If anything, Penrose argues that a computer could not do certain kinds of math, not that they couldnt have a mind . I doubt knowledge of Godel propositions add a mind to a system. But even if we agree with the claims that the mind can do certain things non-computably, it does not follow that consciousness is one of these (remember at the end the book is about consciousness). Now quantum processes are certainly non-computable, so Pernose's claim is that consciousness arises from quantum processes. The problem is that none of this follows from any other discussion before! The quantum is only one non-computable process that could exist in the brain. Also,remember that quantum effects are probably inexistent in such a noisy and hot system like the brain. One even can doubt Penroses claim that there exists mathemathical insight of any kind not explainable in some other way than the quantum. Hammeroff and Nancy Wolf are much better quantum-consciousness theorists, and this review still applies to an extent to Penrose's Shadows of the Mind. I would argue quantum-consciousness is still considered as a real option because it is popular outside the academia. This book was a best seller. But on real scientific terms, it is a no starter. I'm sure some philosophers and physicists might embrace Peroses attempts, but there are still scientists and philosophers that deny the theory of evolution.
It is a good read, and everyone serious in consciousness studies should try to read it, if only for historical reasons. This book is probably a popular science classic allready.
on June 25, 2001
This book seeks to counter intimations from the AI community that the only reason AI has not yet succeeded is the failure to build large enough or complex enough processing machinery. Drawing from a variety of topics in physics and mathematics, Penrose seeks to illustrate many anomalous areas that seem beyond the capability of any kind of artificial reasoning approach currently known, regardless of the amount of sophistication or processing power thrown at them. An "artificial intelligence" that cannot deal with these phenomena, Penrose argues, cannot really lay claim to be what it purports to be -- and we are nowhere near having the tools to do so. Penrose, indeed, is obviously skeptical as to whether it is possible to have such tools.
The book does not end with a grand flourish and "Aha!" The reader is left to draw his or her own conclusions. I found the discussions of specific topics absolutely absorbing, even though I am not a mathematician or physicist. It required a fair amount of focus and concentration on my part to stay with him. It was worth it.
For an educated lay reader with an interest in current science, this is a fascinating and important book.