3. The Missing Science of Consciousness?
A reader who is not convinced by Penrose's Gödelian arguments is left with little reason to accept his claims that physics is noncomputable and that quantum processes are essential to cognition, although these speculations are interesting in their own right. But even if one accepts that human behavior can be accounted for computationally, there is still the question of human consciousness, which after all is Penrose's ultimate target.
Penrose is clear that the puzzle of consciousness is one of his central motivations. Indeed, one reason for his skepticism about AI is that it is so hard to see how the mere enaction of a computation should give rise to an inner subjective life. Why couldn't all the computation go in the dark, without consciousness? So Penrose postulates that we to appeal to physics instead, and suggests that the locus of consciousness may be a quantum gravity process in microtubules. But this seems to suffer from exactly the same problem. Why should quantum processes in microtubules give rise to consciousness, any more than computational processes should? Neither suggestion seems appreciably better off than the other.
Although Penrose's quantum-gravity proposal might at least conceivably help explain certain elements of human behavior (if behavior turned out to be uncomputable, for example), it simply seems to be the wrong sort of thing to explain human consciousness. Indeed, Penrose nowhere claims that it does, and by the end of the book the "Missing Science of Consciousness" seems as far off as it ever was. As things stand, even by the end of Penrose's book, we seem to be left in Penrose's position D: these physical theories leave consciousness entirely unexplained.
This might seem odd, given that Penrose says he embraces position C, but in fact C and D are quite compatible. This is because Penrose's four positions run together a number of separate issues. For convenience, I repeat the positions here:
A: All thinking is computation; in particular, feelings of conscious awareness are evoked merely by the carrying out of appropriate computations.
B: Awareness if a feature of the brain's physical action; and whereas any physical action can be simulated computationally, computational simulation cannot by itself evoke awareness.
C: Appropriate physical action evokes awareness, but this physical action cannot even be properly simulated computationally.
D: Awareness cannot be explained by physical, computational, or any other scientific terms.
Note that A, B, and C all concern how awareness is evoked, but D concerns how awareness is explained. These are two very different issues. To see the contrast, note that almost everybody would accept that the brain evokes awareness - if we were to construct a duplicate brain, there would be conscious experience associated with it. But it is far from clear that a physical description of the brain can explain awareness - many people have argued that given any physical account of brain processes, the question of how those processes evoke conscious experience will be unanswered by the physical account.
To really clarify the positions in the vicinity, we have to distinguish three questions:
(1) What does it take to simulate our physical action?
(2) What does it take to evoke conscious awareness?
(3) What does it take to explain conscious awareness?
In answer to each question, one might say that (a) Computation alone is enough, (b) Physics is enough, but physical features beyond computation are required, or (c) Not even physics is enough. Call these positions C, P, and N. So we have a total of 27 positions, that one might label CCC, CPN, and so on.
Question (1) is the question Penrose is concerned with for most of the book, and the issue that separates B and C above. He argues for position P-- over C--. Descartes might have argued for N--, but few would embrace such a position these days.
Question (2) is the issue at the heart of Searle's Chinese room argument, and the issue that separates A from B and C above. Searle argues for -P- over -C-, and Penrose is clearly sympathetic with this position. Almost everyone would accept that a physical duplicate of me would "evoke" consciousness, so position -N- is not central here.
Question (3) is the central question about the explanation of consciousness (a question that much of my own work is concerned with). Penrose's positions A, B, and C are neutral on this question, but D is solely concerned with it; so in a sense, D is independent of the rest. Many advocates of AI might hold --C, some neurobiologists might hold --P, whereas my own position is --N.
The four positions Penrose describes come down to CC- (A), CP- (B), PP- (C), and --N (D). Penrose seems to think that in arguing for position C (PP-) he is arguing against position D (--N), but it is clear from this analysis that this is not so. In the end, nothing in Penrose's book bears on question (3), which is a pity, though it is certainly understandable. It would be very interesting to hear Penrose's position on just how physical theories might or might not explain human consciousness.
Indeed, one might even combine positions A and D, as I do, embracing CCN. On this position, human-like behavior can be produced computationally, and indeed enacting the right computation will give rise to consciousness, but neither a computational account nor a physical account alone will explain consciousness. It might seem odd that computation should evoke but not explain consciousness, but this is no more odd than the corresponding position that neurophysiology might evoke but not explain consciousness. In either case, consciousness emerges from some underlying basis, but we need a further element in the theory to explain just how and why it emerges.
One can have a lot of fun cataloging positions (Dennett is CCC; Searle may be CPP; Eccles is NNN; Penrose is PPP; I am CCN; some philosophers and neuroscientists are CPN or PPN; note that all these are "non-decreasing" in C->P->N, as we might expect), but this is enough for now. The main point is that Penrose's treatment runs together question (3) with questions (1) and (2), so that in the end the question of how consciousness might be explained is left to one side.
A true science of consciousness will have address all of these questions, and especially question (3). Penrose has produced an enormously enjoyable and challenging book, but it seems to me that for all his hard work, the science of consciousness is still missing.