Roger Penrose tells Kam Patel why he thinks human beings will always be cleverer than computers and why we need a new physics
Most of us at times will come across some aspect of the physical world that puzzles. More often than not the sensation will quickly leave us as more pressing demands on our time arise. But as I talk to Roger Penrose, it becomes clear that, for him, experience of the physical world is something that requires constant, rigorous explanation.
How does he feel his deep understanding of mathematics and physics has influenced his perception of the ordinary, humdrum aspects of the physical world? Inconsistencies bother him a lot, he replies. People, of course, can behave inconsistently - he can appreciate that. But if something in the physical world does not hang together, he worries about it. "My wife worries about people when they are inconsistent, I worry about 'things' when they are," he says laughing.
Light reflected on a ceiling, for instance, can trigger an obsessive search for the source. The light could have been reflected twice or the source may be totally surprising: "A thing like that would get to me until I was satisfied," he says.
Perhaps the most challenging inconsistency exercising the talents of this phenomenally gifted mathematician is the one he sees at the borderline between the theory of quantum mechanics - which describes the behaviour of elementary particles and atoms - and the classical Newtonian laws that explain cause and effect in the observable universe. For the 64-year-old Penrose, Rouse Ball professor of mathematics at Oxford University, a new physics is needed to bridge the two sets of theories. Furthermore, he believes that the new physics, when it arrives, will be a crucial ingredient in any description of higher mental functions and, ultimately, in a description of consciousness itself. It is now more than five years since Sir Roger launched himself into the debate on the nature of human consciousness - the problem of how our awareness of everything, colours, smells, pain . . . is actually formed. In The Emperor's New Mind he argued that consciousness, despite the best efforts of the artificial intelligence community, can never be emulated by computers.
His intervention won support but also attracted criticism - some of it pretty aggressive. In an effort to address "misunderstandings" of Emperor, Penrose wrote a successor, published last year. Shadows of the Mind presents new ideas on the nature of consciousness.
A key feature of Emperor is the application of a theorem first formulated by Kurt Godel in the 1930s - which proves that there are certain truths about a system which cannot be proved solely by acting within that system. One has to step outside the system to appreciate these truths. Penrose has applied Godel's theorem to computational systems, showing, he claims, that there are certain truths that cannot be proven by computation alone but which are "obvious" if one looks at them from outside. The characteristic which enables the human mind to appreciate these truths is understanding, a quality which, he argues, is a function of consciousness incapable of imitation by computer.
He has held the view that there was something beyond computation in physical laws since he first came across Godel's theorem as a postgraduate student at Cambridge in the late 1950s. "Godel's theorem resonated with a feeling that there was something outside computation. It just seems natural and suggests to me that our picture of physics needs to be extended."
Penrose was surprised by the interest Emperor attracted: "I had not expected that. But I also had not anticipated the vehemence of the critical comments from certain quarters - certainly from those who work in artificial intelligence. Some of them were very upset; they felt I hadn't studied their subject particularly thoroughly, that I was coming in from the outside without knowing what could really be done."
I say to him that from an engineering point of view I can see clearly that operating within a system would prevent me from appreciating certain truths associated with its workings. But why assume that my mind should be superior to that system on the basis of my ability to appreciate those truths from outside? The use of the word "superior" is swiftly corrected: "It is a misleading word and I have never used it. I tend to emphasise the word understanding. Okay computers can do a lot of things that are beyond the ability of humans. But how do you know that machine computations are of any use? What is their meaning? You need something quite different - understanding - which tells you what the computations mean, what they are being done for and when they produce answers, what those answers tell you. I am sure that in our brain we do computations but we also have this understanding quality which seems to me to require consciousness."
Is understanding just a different kind of computation? "Well that is what people on the artificial intelligence side would claim," he replies. "But I think that the claim that if you train artificial neural networks long enough they will begin to develop qualities which look like human understanding is simply wrong."
Penrose's often passionate defence of his ideas has led to accusations that he is fighting for the specialness of human beings. He says his position has nothing to do with that. The idea that computers could become aware at some point in the future is one that he would have gone along with in the past but he now believes this argument to be deeply flawed. People who adhere to it have not faced up to the implications of mathematical, logical arguments, he says. It is clear to him that even without employing the Godelian argument, consciousness does not fit in with the universe and the physical world as we understand it: "It is not that I do not want computers to take over the world. Its just that it is unreal, it is not the right theory - it is incorrect. And what I am more interested in is what is correct.
"What I am saying is that there is something in the way the physical world behaves which, if you knew how to harness it perfectly, could, in principle, enable us to build a non-human (maybe not even a biological) device that is genuinely intelligent. From my viewpoint that is possible - we do not know how to do it but it certainly is not going to be via computing. I am not arguing that it is outside some future science."
While writing Shadows, Penrose received correspondence from Stuart Hameroff of the University of Arizona, Tuscon about the possibility that the framework of each living cell, the cytoskeleton, plays an important role in controlling activity at the junctions between nerve cells. It is at these junctions, called synapses, that signals are transferred, usually to another nerve cell. In Shadows, Penrose writes that the classical picture informing our understanding of the brain implies that nerve cells and associated synapses play roles that are essentially similar to those of transistors and wires in the printed circuits of electronic computers.
Penrose was unaware of the cytoskeleton - which is made of protein-like molecules - and was sufficiently intrigued by Hameroff's view that there was computational activity taking place to write back. In his letter, Penrose even said that he thought Hameroff was being "a little conservative" in his theorising - to which Hameroff replied it was the first time anyone had flung such an accusation at him.
Hameroff had focused on a part of the cytoskeleton called microtubules, bundles of tiny tubes organised in a structure with a fan-like cross section. Penrose is not clear as to how much evidence there is for the kind of microtubular computational activity Hameroff is arguing for. He himself however has far greater hopes for microtubules. He thinks they may well be a home for the kind of activity that he believes must occur at the borderline between the quantum and classical worlds. "Here is a place I think where one might take advantage of the new physics. And if consciousness is going to come inside that, these structures are the best candidates for such activity I have ever seen. It was a complete revelation coming across them."
Penrose is claiming that when non-computable activity within the microtubules occurs in "concert" over large areas of the brain it leads to conscious events. "You see what you need from my point of view is something that takes advantage of the borderline between the two theories -- they do not fit together well at all. My view is that one needs a new theory and this is becoming a more general view - it is not just me trying to be awkward."
I ask him what he believes needs to be done to establish the new physics that he is arguing for. Three disciplines in particular would play a vital role, he replies. For physics the task would be to find out what is going on at the quantum-classical borderline. This would involve developing a theory and testing of ideas in experiments: "As far as I can see those experiments are a little outside what can currently be done but not enormously outside." He does not rule out such experimentation being feasible within the next few decades.
For biology the challenge is to find out what is taking place in microtubules, whether there is any evidence for quantum coherence - the concerted, large scale, non-computable action he claims microtubules exhibit. He would like to see biologists study the role of microtubules in governing the strength of signal flowing through synapses and whether microtubules in the brain possess any special features. He says: "Biologists are a little reluctant to take on board quantum mechanics partly because they do not know much about it which is fair enough. But I do think that in the future biology and physics will need to look at each other more."
Psychologists will also have a major role to play in the forging of the new physics through experiments that can, for example, determine the number of neurons involved in conscious events, the timing of conscious events and the relationship of these to microtubules.
What Penrose is arguing for is something that will result in nothing short of a revolution in the way we perceive the physical world. He explains that such shifts are not new. Newtonian physics ruled supreme for 250 years before Einstein came along and with almost no change to the observational consequences provided a totally different explanation - so gravity is not a force anymore, space-time is warped and energy influences the curvature of space-time. All these things completely changed our picture of the world, he says, maintaining that there must be more volte-faces of this kind to come, some of which will have something to do with the structure of quantum mechanics. "We are not going to have a proper picture of how quantum mechanics relates to the classical world without a complete change of worldview. And when these world views have changed enough then perhaps we will begin to see how consciousness fits in."
The arrival of the new physics and, if necessary, successors to it, will also offer the prospect of an accommodation of widely differing views on the nature of the physical world and our place in it. "I do think that the new physics will - at least I hope it will - involve things which reconcile the notions of people who argue from a much more religious perspective . . . they feel quite convinced that there is something there which science will never encompass. And I would agree with them - if by science they mean present-day science. On the other hand there are scientists who say that people who hold such views have nothing to offer. I would hope there is something which one can get from both sides in the new picture that we finally come up with. There is something "mysterious" certainly and as you get deeper into examining how the world holds together you can see that there is something there that has yet to be explained."