The Vertices of Consciousness and the Biology of a Machine

Another way to see this story is that Pygmalian grows increasingly desperate at the distinction between life and his handiwork. He wants the distinction removed, in part to dissolve the artifice from the physical aspect of the statue in order that it correlate with his passionate feelings that treat the object as alive. From a cognitive aspect, Pygmalian confronts the discriminative categorizations of life and art, compounded by the irony of the art work perfectly representing an imagined ideal. The dead image of the ivory woman has an animate counterpart in the sculptor's mind; in the myth, the stone carved image serves as an intermediary in a process that ultimately eliminates the artifice, or more accurately relegates the artifice to memory. The living woman metamorphosed from the stone leaves no residual imprint of her original state. The fact that she was a statue becomes a historical matter of fact.

This myth has poignancy for an understanding of the way that human beings and computers are often portrayed in popular and scholarly discourse. In this brief essay, I would like to summarily explore some aspects of the mythic undercurrents in popular culture and in two influential authors on the subject, Roger Penrose and Daniel Dennett, and then suggest another avenue of computer application for the study of human consciousness.

Movies, videos, and written fiction approach this in myriad ways, often purposefully blurring the distinction in order to heighten the unsettling consequences of a machine who thinks or a human who becomes mechanized. Like the myth, the seemingly miraculous transposition of the machine and the human is also underscored, though unlike the myth the nature of the transformation from artifice to a natural being is invariably messier, with the seamless metamorphosis in the classical tale giving way to a more tortured, conflicting marriage of humanity and the machine. Computers invariably play a part in these marriages, along with other high tech machinery that can fuse, beam, or otherwise transport the human essence to the internal circuitry of a silicon chip.

One can see Pygmalian's handiwork in recent movies of enormous popularity. In the Terminator and Alien series, robots or cyborgs play key roles, and often part of the plot turns on discovering that what seems perfectly human is in fact a machine. These human machines, driven by computers, tend to be unusually powerful physically or have an incredible knowledge base stored internally. In this respect, when discussing the biology of a machine in the context of popular culture -- the lifelike processes and phenomena of a cyborg or android -- superficial descriptions tend to suffice: they are as strong as twenty people combined, or eerily know what twenty very smart people might know. Humanized machines tend to compress quantities of muscle and brain into single entities, often for extremes of good or ill of those around them.

Secondly, and obviously, the correspondence between human beings and machines is not limited to popular culture but occupies some of the most serious research today. Much of this has to do with the nature of artificial intelligence and the ways a computer can model or simulate human thought processes. Not widely known but highly respected is Derek Parfit's philosophical inquiry, Reasons and Persons, which posits important questions for the concept of identity by means of thought experiments that use transport beams that send an individual's molecules from earth to Mars and reassemble the molecules upon arrival. Is the person also reassembled? Is this the same individual who left earth in a stream of electrons?

Better known is Penrose's The Emperor's New Mind, an extended argument against the strong AI position that "mind finds its existence through the embodiment of a sufficiently complex algorithm." (Penrose, 429). Penrose eschews computer/mind analogies, preferring to compare the working of the mind to the growth of a crystal: the activation and deactivation of synapses is like the vast number of possible alternative arrangements of a crystal formation and not a machine with its much more limited arrangements (Penrose, 438).

Part of Penrose's discomfort with the mind/machine analogy is that the use of physical laws in the decipherment of scientific principle is often done through non-algorithmic processes. His own personal introspection and statements by Einstein and Poincare show that innovation and creative thought can be largely non-verbal, highly imagistic and geometrical, and thus not susceptible to algorithmic modeling. Mathematics, a chief source of algorithms, is paradoxical in this regard because many of its hypotheses and subsequent proofs are derived by pictures of "some visual and muscular type," to use Einstein's phrase. It is important to recognize that there are a number of systems that organize reality and interact through a series, however complex, of synaptic firings.

In what seems a direct contrast to Penrose, the variety of functioning systems in the mind is also underscored by Daniel Dennett in his highly regarded Consciousness Explained. The computer, a virtual machine, and software are the salient metaphors Dennett employs for describing the mind's functions. Here the mind/machine analogy receives a most sophisticated treatment. The metaphor conduces to a logical conclusion when Dennett states the desire, and the possibility, of building a robot with consciousness.

The brilliance of Dennett's lucid arguments should not be denied, yet he, too, succumbs to a kind of Pygmalion syndrome that is indicative of the highly mythic content with which we as a culture have imbued our thinking machines. The stereotypes of machines as a physical Terminator or as an intellectual pool of unfathomable depths of knowledge are one dimensional myths. Penrose, while missing some clues as to the flexibility of such things as neural nets and fuzzy algorithms, is probably correct in deriding the strong AI tendency to seek that elusive algorithm that initiates and/or simulates consciousness, a grail quest in the realm of electronic alchemy.

Yet from a certain vantage the seemingly antithetical approaches of Dennett and Penrose are actually arguing with the same cognitive framework that so excited the ancient sculptor. Dennett insists that the marriage of artifice (here advanced AI and computer hardware) and humanity is possible and desirable. Penrose adamantly draws the boundary between the two and argues that the transformation of one into the other is impossible and rather foolish as an enterprise.

While I personally find Dennett's arguments much more convincing and exciting, one must ask what, precisely, will his conscious robot accomplish, other than final proof of the misdirection of Cartesian dualism? If we could build a robot who thinks, would the attributes of longevity, dependability, strength, compliance, predictability, and accumulated knowledge be the end results of such a design? If a robot were truly like a human, these qualities would be genuinely compromised. Is this more like building superhuman beings? If so, why would machine parts driven by a human like consciousness and a genetically engineered species of living beings that would look and act indistinguishably from a perfectly sculpted thinking machine tend to be abhorrent? Should these slightly less than parallel lines of thinking converge imaginatively as a truly human like machine and as a biologically engineered product with completely organic parts, the only way to distinguish them would be to know their history. The memory of their origins, the memory of their original categorization as it were, would be needed to distinguish the superficial reality of seeming flesh and thinking mind. Another method of determining their nature, often used in cinema, is to cut them in half.

Instead of building in our image, we should build to better understand precisely the workings of the mind that conduce to such an image. It is time we rethink our imaginative correspondence to humans and machines, indeed to scrap the notion of building something human altogether. The distinction between being human and a machine should rather be stressed. More effort should be pointed toward a genuine fifth generation application rather than returning to what are essentially mythic permutations of Babbage's first generational dream. Penrose is essentially incorrect in rejecting a correlation between the mind and the machine; there is indeed a correspondence, but not necessarily one of perfect emulation or a precise categorical match. The messiness, often quite literal, of combining humans and machines in science fiction is closer to the mark, though again this is superficially handled for the most part .

Computers should be built to model the one distinctive feature of humanity, consciousness (not intelligence), and to do it as a machine, for the quite simple fact that we are only now learning what consciousness is, and thereby learning who and what we are. We have been staring at our own exteriors reflected too long, and it should be remembered that Narcissus, similarly transfixed, died not because of the beauty he saw reflected and could not attain, but because he did not know himself, did not know it was his face in the reflecting forest pool. He died broken upon the realization that the image was he, a poignant instance of the tragic side of Socrates' dictum and a paradoxical twist on the frustrations of Pygmalian.

How might we model consciousness? Two aspects are important: one is that consciousness has intersections or vertices, and the other is that information stored and manipulated in the mind is done so by biological principles, not mathematical or necessarily logical ones. It is this biology of information that a machine needs to emulate.

Some suggestive clues concerning the vertices of consciousness come from experiments with the famous robot Shakey in the 1960s. Shakey was programmed to recognize objects in a room, but not by identifying the objects as whole entities. Rather, Shakey could discern a square because of the vertices a square uniquely has at the point where its angles meet. This held similarly for triangles and other objects. Shakey could generalize by recognizing a small part, the part where planes of the object intersected.

It is also becoming clear from other studies that consciousness has a number of planes or facets and that they are quite dynamically interactive. The new understanding of consciousness is revolutionary in its way, as scientists like George Lakoff have eloquently explained. From the classical models of the mind as an assembly of atomistic building blocks of conceptual categories that are linked by a calculable logic has come a dynamic model of much more flexible categories that structure knowledge in an evolving way. Intuition, creative thinking, and emotions are inextricable from recent tenets of cognitive studies. In part, this understanding represents a movement away from a mathematical/logical model to a biological one, wherein mind and brain are one and thereby more dynamic. There is no one watching in our heads; a thread of nerve and electrons is mind or the interaction of these threads produces mind.

A machine does not have to be biological to emulate the biology of information and knowledge structuring of consciousness any more than it needs a heart to correlate blood samples. Biology in this sense allows for more random cross linking and fuzzy kinds of associations, and this fuzziness is most apparent in the vertices of two mental systems of categorization and structuring of knowledge: words and images, the bugbear of Penrose's exposition.

The confrontation of these systems in the mind, or as mind, can pose a fascinating avenue in the development of thinking machines. Discarding Pygmalian, one can move closer toward the firelit cave of Plato's metaphor, wherein mere shadows of reality flicker on the wall. Shadows are just fine as a start, and one need not look far to see the implications of categorical shadows colliding as a process of consciousness.

Formulation of principles of quantum mechanics or three dimensional global weather modeling can be taken to show the immense power of the contemporary computer. To understand the subtle complexity of consciousness and what a machine today cannot do, one needs to turn to the common riddle. Few are as famous as the one uttered by the Sphinx that Oedipus solved: what walks on four legs when born, two in the middle of life, and three at the end of life? If analyzed linguistically, the literal definitions of these words pose no problem. There is nothing odd about the syntax or about the grammar of the question.

The puzzle arises from the non-literal aspect of the words, the connotations, suggestions, allusions, and images formed: how can something transform like this? Once we know the answer, a human being, the non-literal references become clear. This riddle, like most riddles, plays upon and against our categories of knowledge structures. Almost always a living entity begins with the same number of legs it is born with. But legs here is meant in a connotative way, downward extensions that support the body in its movement, so arms are legs and a crutch or cane is a leg.

The riddle is highly imagistic, in that part of its confusion lies in trying to picture this animal. The odd number progression from 4 to 2 to 3 also is not intuitive and is perplexing. One of the tendencies when confronted with this difficult image is to resort to myth, where extraordinary transformations like this might more likely occur. The wonderful irony is that the creature in question is anything but mythic. Another riddle which I am fond of retelling comes from Renaissance England: I am within as white as snow, without as greene as hearbs that grow; I am higher than a house, and yet am lesser than a mouse.

The answer is walnut. Like the riddle of the Sphinx the grammar and syntax pose no problems. The standard definition of each word is not challenged in any significant way. Bringing the four short sentences together to form a single category produces the riddle. The answer shows that the allusion white as snow must mean the flower; that greene is the leaf; the tree itself can be higher than a house and the seed smaller than a mouse. We are nonetheless accustomed to think of these stages in a linear way. Walnut probably exists in all these forms simultaneously on the planet, and perhaps at times on a single tree, but we think of the vast size differences and colors in an antithetical, not combinatorial fashion, usually, and the riddle plays precisely upon this linear, organizing scheme we have of reality, particularly in its use of the present tense existential verb. Again, without knowing the answer, the object is impossible to picture.

Within these small riddles reside the margins of fundamentally human thoughts. Far from the crushing grip of cybernauts or the bright red rectangles of HAL's exhaustive memory banks is the riddle of consciousness itself. These margins are riddles because of the way consciousness works: language is not nearly as precise as linguistic computer analysis tells us, and images in the mind come in a variety of fluctuating shapes and degrees of clarity that defy the predictability, however tenuous, of the jet stream. When combined they produce an extravagance of possibilities with something akin to the logic of chaos.

We could use the assistance of computers. It may be beyond our grasp to build a machine to capture this process, so enamored are we of physical replication and the quantity of thought. We seek too often for the blinding light of Plato's illumination, where in truth the essence of our intellectual life is shadow. If dream itself is but a shadow, so are the foundations of our knowledge, and where those shadows touch lies understanding and, perhaps, wisdom. Because of the flux and dynamism of the biology of information, truth can only evolve. The next generation of computers may serve figuratively as the box of fire that casts our internal shadows upon the cave, where we can watch them flicker and transform, and thereby know of the complex community we share within the life of the mind.

Robert L. Nadeau, Mind, Machines, and Human Consciousness (Chicago: Contemporary Books, 1991)

Roger Penrose, The Emperor's New Mind: Concerning Computers, Minds, and the Laws of Physics (New York: Oxford University Press, 1989).