The problem with zombies

April 5, 1996

Daniel Dennett's stimulating Consciousness Explained failed to live up to its title. Now another philosopher, David Chalmers, offers a new theory. Roger Penrose assesses it scientifically

David Chalmers presents his views on consciousness and its relation to the physical world very much from a philosopher's perspective. Since I find many of the issues that tend to occupy philosophers' attentions hard to relate to, I am approaching his work more as a contribution to our scientific understanding of the elusive nature of consciousness, than as a philosophical discourse.

Yet, scientists have much to gain from the philosopher's input. There are many confusions to be clarified at the very foundations of our physical pictures. The role of consciousness, in relation to the physical world, provides perhaps the deepest of these potential confusions. Chalmers is a philosopher of distinction who has thought long and hard on these matters and is well-versed in most of the fundamental issues underlying present-day physical theories. I therefore started with high hopes that his own insights could shed important clarifying light on these central issues. In this, I felt somewhat disappointed. However, I believe that there are valuable arguments given here which may contribute importantly to our final understanding of the puzzle, although perhaps not in the way that Chalmers intended.

Much of the book is in a philosopher's style that I find uncongenial. Unfamiliar words are introduced in order to make what seem to be hair-splitting distinctions. For example, numerous subtly different notions of consciousness are discussed at length. Can one be "aware" of a sensation without "consciously" noticing it? This relates to the philosopher's conundrum of a continuous noise that suddenly stops: one may become conscious of the noise only after it stops. (But is this not merely a reflection of the time-delays that can be involved in conscious awareness that we know are present from the important experiments of Benjamin Libet and his colleagues?) Definitions are often confusing to an outsider such as myself because they are frequently given only by example. I never grasped the sense of "naturally possible" - as applied to a randomly acting monkey typing Hamlet, but not to a persisting cubic mile of uranium 235. (Both are merely matters of probability.) There is also much analysis of the statement "water is H2O". (Believing that steam, not water, is H2O, I found this additionally confusing.)

"Zombies" - who act exactly like conscious human beings while being entirely unconscious - occur frequently. It is argued that such beings are conceivable, and the conclusion is drawn that consciousness cannot be a physical phenomenon. Why not more reasonably conclude that an unconscious physical being must behave differently from a conscious one, consciousness being a physical phenomenon? (Just because Chalmers can imagine a zombie does not make it possible; I can "imagine" a counter-example to Fermat's last theorem.)

Much faith is placed in the type of argument which depends upon the assumed theoretical possibility of successively replacing every individual neuron in a conscious person's brain by "a silicon chip that performs precisely the same local function as the neuron". Using ingenious new arguments (concerning "fading" and "dancing" qualia), Chalmers persuasively deduces that a robot whose silicon chips are wired in precisely the same way as are the neurons in his own brain would have just the same conscious experiences as himself.

Chalmers is much less convincing in deducing that John Searle's Chinese room could actually experience a "redness" sensation. Accordingly, while I side with Chalmers on his first argument, I must support Searle on the second - which seems to lead us to a contradiction: both arguments require the assumption that such function-preserving neuron replacements are possible; hence they are not possible! Previously, I had depended upon arguments from Godel's theorem to arrive at such a conclusion - but we now see that it also follows from this completely different line of reasoning. (Chalmers barely refers to the Goedelian case, dismissing it in half a page with an incorrect argument. Also, his interpretation of a finite-state Turing machine as an infinite "combinatorial-state automaton" is inappropriate, for reasons I cannot go into here.)

Perhaps a fundamentally non-computable physical input occurs at the neuron's cytoskeletal level (as Hameroff and I have proposed) - depending upon a presently unknown physics at the quantum/classical borderline. Chalmers, however, is dismissive of the possibility that our present-day quantum theory needs fundamental change. Instead, he is driven to the "many-worlds" (or "many-minds") Everett interpretation which he admits is "almost impossible to believe". It is here that his arguments become least credible (and if he really believes them, he should go over the entire reasoning of his book all over again, in the light of his changed perspective).

Perhaps it demands an unreasonable boldness to accept that our present-day quantum physics requires revolutionary change - as with Einstein's overturning of Newton's superbly accurate gravitational theory. Yet I believe that such a change is necessary, and the strong artificial intelligence/many-worlds deductions that Chalmers feels driven to, are unwarranted. But he is right to stress the inadequacy of "conventional" scientific approaches to consciousness, and his logic is normally impeccable. I believe that there is much of lasting value in his book, despite my profound disagreement with his final conclusions.

Sir Roger Penrose is Rouse Ball professor of mathematics, University of Oxford.

Register to continue

Why register?

  • Registration is free and only takes a moment
  • Once registered, you can read 3 articles a month
  • Sign up for our newsletter
Register
Please Login or Register to read this article.

Sponsored