"Shadows of the Mind: A Search for the Missing Science of Consciousness." By Roger Penrose, Oxford University Press 457 pp.; $25

"The Engine of Reason, the Seat of the Soul: A Philosophical Journey into the Brain." By Paul M. Churchland, MIT Press 329 pp.; $29.95

Possibly the most challenging and pervasive source of problems in the whole of philosophy. Our own consciousness seems to be the most basic fact confronting us, yet it is almost impossible to say what consciousness is.

So begins the entry on consciousness in "The Oxford Dictionary of Philosophy," by Simon Blackburn (Oxford University Press, 1994). Few would dispute the author's judgment that the nature of consciousness is a particularly daunting puzzle. It is also, however, a puzzle that a number of thinkers in a variety of disciplines claim to have solved--in principle, at least. Witness the title of Daniel C. Dennett's widely praised book, "Consciousness Explained" (Little, Brown, 1992).

There is no scholarly consensus here, and some prominent skeptics--Jerry Fodor and Colin McGinn, for example--have suggested that the problem of consciousness is by its very nature unlikely ever to be solved. Still, the problem is attracting an extraordinary amount of attention. Dennett's book is just one of a large shelfful of recent studies of consciousness, beginning with "The Emperor's New Mind: Concerning Computers, Minds, and the Laws of Physics," by the mathematician and physicist Roger Penrose (Oxford University Press, 1989)--a book that appeared on bestseller lists at university campuses.1 In addition to the books intended for general audiences, there has been a steady flow of more specialized works. And in 1994, "The Journal of Consciousness Studies" began publication.

The surge of interest in consciousness has been driven in part by developments in two areas: neuroscience and computational models of the mind. Claims based on advances in these areas tend to set the agenda for consciousness studies--even in the case of thinkers who reject many of those claims (Roger Penrose and John Searle, for example). Two new books--"Penrose's Shadows of the Mind: A Search for the Missing Science of Consciousness" and Paul M. Churchland's "The Engine of Reason, the Seat of the Soul: A Philosophical Journey into the Brain"--suggest the range of current attempts to establish the physical basis for human consciousness and underline the significance of that search.

"Shadows of the Mind" is a sequel to "The Emperor's New Mind." In part 1, "Why We Need New Physics to Understand the Mind," Penrose argues as he did in the first volume that human mathematical understanding is inherently noncomputational, meaning that it could not, in principle, be simulated in its completeness by a computer. In part 2, "What New Physics We Need to Understand the Mind," he speculates about the possibilities for developing a scientific understanding of consciousness, an understanding he argues is impossible in terms of current science.

The entirety of part 1 is really an extended development of a single argument, based on the famous mathematical result known as Godel's Theorem. (The discussion inevitably becomes technical at points. Penrose tries to simplify it and to avoid relying on prior specialized knowledge, but the reader who is not fairly knowledgeable about science and mathematics is likely to find the going tough in places. Penrose encourages such a reader to skip the mathematics when it becomes impenetrable and to omit whole sections not needed for the basic understanding of the argument.) Godel's Theorem asserts that every sufficiently complex, consistent, mathematical system is "incomplete" in the following sense: There is a true statement (the "Godel sentence") that cannot be proved within the system, yet the sentence in question can be shown to be true, by the special methods devised by Godel himself.

This shows, according to Penrose, that human mathematical understanding cannot be represented by such a formal system. For the system cannot prove that its own Godel sentence is true, whereas the human mathematician, using mathematical insight, can prove this. Thus the human can do something the system itself is unable to perform. Of course, the point is not that mathematical understanding is unique in being noncomputational; rather, all true understanding is said to be noncomputational, but the rigorous nature of mathematical thought is what makes it possible to prove this conclusively in the area of mathematics.

Is Penrose right about this? I have no doubt that human understanding is noncomputational, but does Penrose succeed in proving it? His argument is complex and deserves a more thorough evaluation than is possible in this review. It seems to me, however, that at best what he has proved is the following: If human mathematical reasoning can be captured in a computational system, then the system is one no human being can grasp. (A human mathematician cannot prove her own Godel sentence.)

This result, if accepted, might be somewhat discouraging to those pursuing the goal of "artificial intelligence." If they are unable to grasp the "human computational system" in its completeness, they will also be unable to construct an artificial system that possesses all the mathematical capabilities of human beings. It is not clear, however, that those who favor a computational theory of the mind need find the result unacceptable; it follows (if I am right about this) that Penrose has not proved that human consciousness is noncomputational. But readers interested in this type of argument should consult "Shadows of the Mind" for themselves.

In the second part of the book, Penrose takes the reader on an impressive tour of contemporary physics, arriving at the conclusion that no currently known physical theory can account for the noncomputational nature of the mind. Yet he insists that he does want a physical explanation of the basis of consciousness, and so he speculates about where in physics and biology such an explanation might be found.

After chapters on "Does Mind Have a Place in Classical Physics?" and "Structure of the Quantum World," we reach the chapter on "Quantum Theory and the Brain," where Penrose's positive suggestions are presented; a chapter entitled "Implications?" ends the book. Perhaps most significant is Penrose's recognition that, if consciousness is something that in the end has a physical explanation, the physics involved--the account of the nature of what we call "matter"--must be considerably different from physics as we know it at present.

Paul Churchland, professor of philosophy at the University of California, San Diego, agrees with Penrose in seeking the physical basis of human mentality, but in most other respects their books stand in contrast. Penrose argues at length that human consciousness is noncomputational; according to Churchland, on the other hand, it is computational through and through--though admittedly in a different sense of "computational" than Penrose has primarily in mind.

Penrose argues that existing science offers no adequate basis for understanding the nature of consciousness and proceeds to probe the outer limits and unsolved problems of science in search of such a basis. Churchland, in contrast, seems quite satisfied with current science, so far as foundational understanding is concerned, and invites the reader to share his excitement in exploring the various dimensions of contemporary "cognitive science." Last but perhaps not least, there is the difference in style. Penrose's book, though well written, will prove a challenge in places to all but the most advanced readers. Churchland, on the other hand, has written Engine of Reason in a breezy, engaging style that will appeal to readers from undergraduates on up.

"How does the brain work? How does it sustain a thinking, feeling, dreaming self? How does it sustain a self-conscious person? New results from neuroscience and recent work with artificial neural networks together suggest a unified set of answers to these questions." The first part of the book, entitled "The Little Computer that Could: The Biological Brain," aims to "make those scientific developments available, in a lucid and pictorial form, to the general reading public." (The book is generously illustrated, with both line drawings and photographs. There is even a simple stereo viewer, stored in a pocket inside the back cover, to enable the reader to appreciate stereo images that occur in two chapters.) The developments are indeed impressive and should prove fascinating to readers irrespective of their previous theoretical orientation.

The substance of the first part of the book lies in information about the brain gleaned from neuroscience and reports on the successes that have been achieved in simulating brain function on artificial "neural networks." A key idea here is that of parallel distributed processing (pdp), in which one pattern is transformed into another by "passing it through a large configuration of synaptic connections." This is a different style of computing than the more familiar serial processing found in conventional computers, in which a computer performs specified logical operations one at a time according to a predetermined sequence (the "program"). Neural networks are not "programmed" to perform their tasks; rather, they "learn" to perform them by being "trained" on a set of examples and gradually adjusting the "weights" of the "synaptic connections" between the artificial "neurons" in the computer. An interesting feature of this process is that a neural net can develop impressive abilities (such as the ability to recognize particular faces from a set on which it has been trained) without human beings having any detailed understanding of the processing involved.

Neural nets have been remarkably successful on many tasks, especially the sorts of tasks, such as perceptual recognition, that have proved intractable for traditional computers using linear processing. A neural net designed and trained to translate English-language text into spoken phonemes (thus enabling a computer to "read aloud") achieved in a fairly brief period a proficiency comparable to that of a commercial "linear" program that took several man-years of programmer time to prepare. Other neural nets have developed proficiency for stereoscopic vision and the recognition of human facial expressions, among many other examples. There is, furthermore, considerable evidence that the methods used by the neural nets approximate those employed by the brain itself in solving similar problems. At a minimum, pdp using neural nets is much more similar to actual brain function than is traditional linear processing.

The second part of Churchland's book, entitled "Exploring the Consequences: Philosophical, Scientific, Social, and Personal," is where the controversial issues are fully joined. Perhaps the most important chapter here is chapter 8, "The Puzzle of Consciousness." Unfortunately, this is also the most disappointing chapter in the book.

A central aim of the chapter is to refute the claim of philosophers such as Thomas Nagel, Frank Jackson, and John Searle that the qualitative nature of conscious experience is something that cannot be explained in terms of physical science. In a famous article, Nagel asks "What is it like to be a bat?" His point is that there must indeed be "something it is like to be" a bat; we don't seriously doubt that bats do have subjective conscious experiences. But just what it is like to be a bat is quite unknown to humans; we lack a particular sort of sensory experience--the capacity to locate objects in our environment by bouncing sound waves off them--that is crucial for bats. Furthermore--and this is the essential point--no amount of scientific knowledge concerning the structure and function of a bat's brain can tell us what it is like to be one.

Churchland's answer to this argument contains serious misstatements about the philosophers he is criticizing. He implies, wrongly, that Nagel and Jackson are Cartesian dualists. And he criticizes Searle, in a thoroughly patronizing way, for adhering to a doctrine Searle himself explicitly repudiates, namely, the "infallibility of introspection … [which] has by now been so thoroughly discredited that it is plain curious to find a philosopher of Searle's prominence still clinging to it." Curious indeed--especially since Searle doesn't!

More important, however, Churchland's answer to the argument simply misses the point. He agrees that we have a particular way of knowing our own conscious experiences that we do not have in the case of other creatures, such as bats, or even for other human beings.

"Each individual gets information about the moving tapestry of his own sensory activity via a specific set of neuronal pathways that only he has." But, he says, the fact that one has a way of knowing one's own conscious states that is unique to oneself in no way shows that what is known in this way is different from what someone else knows by the ordinary methods of third-person observation. After all, each of us has a special way of knowing--through the proprioceptive system--about the positions of our own limbs, but this does not imply that the object of knowledge is different for oneself than it is for others.

This, of course, is true, but it misses the point. Nagel's point--and Jackson's, and Searle's--is that the experiential quality of the conscious experience lies beyond the reach of physical science, something that obviously is not true of the positioning of one's limbs. It seems to me that Nagel, Jackson, and Searle are just right about this, and this fact, while not necessarily fatal to materialism, is one with which materialistic philosophers like Churchland have to deal. Churchland, however, manages to ignore it.

And in the next chapter, "Could an Electronic Machine Be Conscious?," he contends that a computer that functions in the same way as a brain could in principle be built using pdp and neural nets, and that such a computer, just because it functions this way, would be a conscious machine. There is no need to go looking, with Penrose, on the murky frontiers of science for the basis of consciousness; all we need is the right combination of copper wires and silicon chips, and your "friendly neighborhood robot" becomes a reality.

There is much, much more in the book that invites discussion. Churchland rejects--apparently with good reason--Noam Chomsky's contention that the human brain contains a special "language organ," hard-wired with the rules of universal grammar. And he criticizes Daniel Dennett's theory of consciousness, on the grounds that Dennett has not properly appreciated the advantages of pdp over traditional serial processing.

Readers of BOOKS & CULTURE may be especially interested in Churchland's attitude toward religion. In a word, it is dismissive; the most significant thing he says about religion is that religious ethics, because of its traditionalism, keeps us from learning from our mistakes and making moral progress. Presumably Churchland is delighted that the lessened influence of religion at present has opened the way for a much greater latitude in moral experimentation.

As has been noted, the one thing Penrose and Churchland conspicuously agree on is the need to find the physical basis of human consciousness. This could be a mistake--but then again, perhaps not. There is a large amount of evidence, some of it described by Churchland, that the functioning of the human mind is highly dependent, in very specific ways, on the functioning of the brain and nervous system. Mind-body dualists, who think the mind or soul is not fundamentally dependent on the brain, owe us a plausible account of these functional dependencies--an account that, so far as I know, is not yet forthcoming.

On the other hand, the view that the mind is somehow produced or generated by the brain is not in conflict with any essential Christian doctrine, including the belief in eternal life. It may be that we need to take more seriously than has generally been done the truth that we are created from the dust of the earth.

Beyond this point of agreement, Penrose and Churchland diverge radically. Penrose's speculations, though in no way conclusive, should on the whole be welcome to Christian thinkers: he recognizes the profoundly mysterious nature of mind and consciousness and eschews simplistic, reductive solutions. Churchland, on the other hand, is reductivist to a degree that even many philosophical naturalists find implausible; he actually thinks we will eventually replace our present ways of talking about the mind, in terms of beliefs, desires, intentions, and the like, with talk in the language of neuroscience. (This is his "eliminative materialism.")

At present, leadership in the philosophy of mind is largely, if not exclusively, in the hands of naturalists and materialists. There is need and, I believe, also a genuine opportunity for serious, constructive work by Christian philosophers in this vital field of philosophy. An excellent (and extremely readable) book on the subject written from a Christian standpoint is Charles Taliaferro's "Consciousness and the Mind of God" (Cambridge University Press, 1994); let us hope it will be followed by many more.

1. Among the other books on that shelf are "The Mind Matters: Consciousness and Choice in a Quantum World," by David Hodgson (Clarendon Press/Oxford University Press, 1991); "The Problem of Consciousness," by Colin McGinn (Blackwell, 1991); "Bright Air, Brilliant Fire: On the Matter of Mind," by Gerald M. Edelman (Basic, 1992); "Consciousness Reconsidered," by Owen Flanagan (MIT Press, 1992); "A History of the Mind: Evolution and the Birth of Consciousness," by Nicholas Humphrey (Simon & Schuster, 1992); The "Rediscovery of Mind," by John Searle (MIT Press, 1992); "The Astonishing Hypothesis: The Scientific Search for the Soul," by Francis Crick (Scribners, 1994); "Descartes' Error: Emotion, Reason, and the Human Brain," by Antonio R. Damasio (Grosset/Putnam, 1994); and "Mental Reality," by Galen Strawson (MIT Press, 1994).

Copyright (c) 1995 Christianity Today, Inc./BOOKS AND CULTURE Review

bcjan96bccurrmrj6B1020614v