Abstract. We refute the arguments made by Roger Penrose that computability considerations suggest the impossibility of consciousness based on the current digital architecture of computers and that better theories of quantum collapse and quantum gravity are needed to understand the consciousness. Instead we argue that for a "materialistic" theory of consciousness it is probably necessary to consider material processes both inside and outside the "border of the body" of a conscious entity (i.e. that the actual body of a conscious entity is not restricted by its borders) and that it is likely that a reasonably advanced theory of consciousness is necessary for physically meaningful theories of quantum collapse and quantum gravity.
Much more can be written about books "The Emperor's New Mind: Concerning Computers, Minds, and the Laws of Physics" and "Shadows of the Mind: A Search for the Missing Science of Consciousness" by the well-known mathematician and scientist Roger Penrose, than it is possible to fit within one essay. Extensive discussion of his books can be found here (alternative URL). This discussion contains a lot of the detailed critique of Penrose's arguments and his attempts at their rebuttal.
So what is the use for one more review? I think that Penrose's books are quite brilliant, and thinking about them helps us to achieve a much better understanding of problems of consciousness and AI. At the same time I also think that Penrose is quite wrong on all main counts. I'll try here to outline only the most important errors in his thinking, but at the same time to also extract the key insights one might derive from this.
The first main argument of Penrose is that the combination of Goedel's incompleteness theorem and apparent ability of mathematicians to uncover the truth about the Platonic world of mathematical objects means that any AI based on the current architecture of digital computers will have essential deficiencies compared to humans (and, in fact, would not have consciousness).
The second main argument of Penrose is that the phenomenon of quantum collapse is the only source of incomputability in the world and the only place to hide the freedom of will. Penrose correctly assumes that we do not have a good understanding of quantum collapse, and that the progress in this understanding is closely linked to the progress in the theory of quantum gravity and unified quantum field theory which would include gravity. His conclusion is that the progress in these fields is a necessary prerequisite to the understanding the nature of consciousness.
Computers and humans are not closed systems, and their physical borders are somewhat artificial.
We start with the first argument of Penrose. The bulk of the related discussion addresses a very scholastic question: whether a computer (or a human for that matter) held essentially in a sensory isolation can develop and/or maintain intellect. However, the real question is whether a computer in a real environment can develop and maintain intellect and consciousness. Penrose basically attributes the "non-computational effect" to the quantum world, and any real computer is immersed in such a world, and can take full advantage of the quantum effects taking place in that world including the world's feedback to the signals from the computer.
This shows that no argument from the classical theory of computability can be applicable to resolve the question of the possibility of (at least) human-level consciousness based on the current digital architecture. Digital computers essentially supress quantum indeterminism inside their circuits, but still can make use of such indeterminism outside those circuits (on even inside their own wires on the smaller scale, if need be eventually). While it is possible that such a suppression might for some currently mysterious reason preclude the computer from gaining such things as subjective experience, I do not currently see how a good argument to this effect can be put together.
The moral of this story is that both computers and people are not closed systems, and that environment is essential for their functioning. Pushing this line of thinking further, we might be willing to conclude that the physical borders of a human body or a computer body are somewhat artificial. Perhaps, it is more appropriate to consider a much larger environment, or conceptually even the whole universe, as such a body, in which the "borders", and what is "inside" and "outside" them are parts of a single mechanism.
This shows that a lot of misconceptions in and about materialistic theories of consciousness comes from the implicit postulate that the consciousness (subjective experience) is only related to the processes within the borders of the body. In fact, it is likely that there is no correct way to even formulate such a statement (even in "Matrix" there is an exchange of signals between the body and the rest of the universe).
Relationship between quantum theory and theory of consciousness.
I think Penrose correctly points out that on the ultimate, philosophical level there must be a deep relationship between quantum theory and theory of consciousness. Indeed, it seems that the room for the "honest" freedom of will at the "micro-level" is likely to be found only within future developments of quantum theory and, in particular, the theory of quantum collapse.
For example, the evolutionary argument that freedom of will is a real thing and not an "illusion" was discovered by many people independently (it is likely that William James was the first, and then Popper and Eccles did that independently, and so on). The idea is that if we postulate that the complex subjective experience is a result of evolution, it could only be so, if it were adaptive, and it could not be adaptive, if there were no way for the subjective experience to influence physical reality. Hence subjective experience is not an epiphenomenon, and freedom of will is not an illusion.
We here assume (I believe together with Penrose) the identity theory of matter and consciousness, or, more precisely, that matter and consciousness are indivisible parts of the same "thing", just like electricity, magnetism, and light are indivisible parts of the same "thing" (electromagnetism). Then the unified theory of matter and consciousness might be considered as an ultimate goal of science, and, in some sense, as a generalization of the unified theory of matter, as Penrose seems to suggest. So it is natural to assert close relationships between quantum theory of collapse (and, perhaps, quantum gravity) and theory of consciousness.
I diverge from Penrose, when he states that there can be no theory of consciouness without the good theory of quantum gravity and quantum collapse. The somewhat unexpected conclusion I am about to make is dual: perhaps, we were failing so miserably to produce decent theories of quantum gravity and quantum collapse so far, because the physically meaningful versions of these theories might be impossible without the inclusion of the theory of consciousness!
In fact, the theory of consciousness to me is first of all the theory of subjective realm (I am not a functionalist by any means). It is quite likely that for a quite wide range of subjective phenomena methods of classical physics would be sufficient for quite excellent approximations, just as they were sufficient in the objective realm. On the other hand, it is the more exotic and less verified subjective experiences (many of them are certainly quite real, because there is enough first hand experiences from credible sources), which are likely to need quantum explanations.
However, if quantum collapse is the ultimate low-level expression of free will (and the resulting distributions are basically the results of "census" of population of particles), then it is not surprising that we cannot find anything better than questionable probabilistic models without taking this free will into account. And if quantum gravity and unified quantum field theory are supposed to constitute "the general theory of everything", they should include the theory of consciousness, because the consciousness is part of that "everything".
So my take is different from the take of Penrose. It is not that we need quantum gravity for a decent theory of consciousness (although we might need it at some later stage of theory of consciosness). It is more likely that we will need a sufficiently general theory of consciousness to obtain a physically meaningful theory of quantum gravity.
Mishka --- February 2002
Copying of this and my other papers on the science of consciousness is allowed free of charge, provided that the texts and this notice are unaltered, and that no further restrictions on the subsequent free redistribution are imposedBack to Mishka's home page