Wonder Tissue

In his excellent book on Indian street magic, Net of Magic: Wonders and Deceptions in India, Lee Siegel (1991) writes,

“I’m writing a book on magic,” I explain, and I’m asked, “Real magic?” By real magic people mean miracles, thaumaturgical acts, and supernatural powers. “No,” I answer: “Conjuring tricks, not real magic.” Real magic, in other words, refers to the magic that is not real, while the magic that is real, that can actually be done, is not real magic. [p. 425]

“Real magic” is—by definition, you might say—miraculous, a violation of the laws of nature. Many people want to believe in real magic. When the Amazing Randi, magician, skeptic, and ghost-buster supreme, duplicates the tricks of self-styled psychics such as Uri Geller, he is showing that the startling effects are not real magic; they are conjuring. But this doesn’t convince some people. At the question-and-answer session following a performance in Winnipeg many years ago, a member of the audience charged Randi with a double deception: he was just as much a real psychic as Geller—he just pretended to be a mere conjurer exposing Geller’s duplicity so he could ride Geller’s more famous coattails to fame and fortune! It’s hard to rebut that weird challenge, except by teaching everybody in the audience exactly how the tricks are done, which Randi, honoring the tradition of magicians around the world, is reluctant to do. (Penn and Teller have pioneered routines that do expose the secret mechanics of magic tricks, and after enduring condemnation by some of their fellow magicians, they have shown how the traditional taboo can be violated without spoiling the magic show.)

A similar eagerness to believe in real magic afflicts many people when the topic is the relationship between the mind and the brain. Some people, including not a few neuroscientists and psychologists—and philosophers—are at least subliminally attracted to the idea that somehow or other the dynamic properties of neural tissue can do something you might call miraculous, something that harnesses hidden forces undreamed of by science. Maybe they are right, but we mustn’t assume that they are from the outset. The rule has to be: no wonder tissue!

Here is something we know with well-nigh perfect certainty: nothing physically inexplicable plays a role in any computer program, no heretofore unimagined force fields, no mysterious quantum shenanigans, no élan vital. There is certainly no wonder tissue in any computer. We know exactly how the basic tasks are accomplished in computers, and how they can be composed into more and more complex tasks, and we can explain these constructed competences with no residual mystery. So although the virtuosity of today’s computers continues to amaze us, the computers themselves, as machines, are as mundane as can-openers. Lots of prestidigitation, but no “real magic.”

This matters because any time we can make a computer do something that has seemed miraculous, we have a proof that it can be done without wonder tissue. Maybe the brain does it another way, maybe even with wonder tissue (maybe Randi is a real psychic, just like Geller!), but we have no good reason to believe it. Computers thus play an important role as demystifiers, and that is a good reason to insist on developing computer models of anything we are trying to understand, whether it be hurricanes or housing bubbles or HIV or human consciousness.

The term wonder tissue is a thinking tool along the lines of a policeman’s billy club: you use it to chastise, to persuade others not to engage in illicit theorizing. And, like a billy club, it can be abused. It is a special attachment for the thinking tool Occam’s Razor and thus enforces a certain scientific conservatism, which can be myopic. I owe my favorite example of this to William Bateson, one of the fathers of modern genetics, and here’s what he had to say not so long ago, in 1916:

The properties of living things are in some way attached to a material basis, perhaps in some special degree to nuclear chromatin [chromosomes]; and yet it is inconceivable that particles of chromatin or of any other substance, however complex, can possess those powers which must be assigned to our factors or gens [genes]. The supposition that particles of chromatin, indistinguishable from each other and indeed almost homogeneous under any known test, can by their material nature confer all the properties of life surpasses the range of even the most convinced materialism. [p. 91]

He just could not imagine DNA. The idea that there might be three billion base-pairs in a double helix inside every human cell was simply not within the scope of his imagination. Fortunately, other biologists didn’t share Bateson’s pessimism, and they went on to discover just how the apparently miraculous feat of transmitting genetic information from generation to generation is achieved by some pretty fantastic molecules. But along that path of discovery, they held themselves to the rule: no wonder tissue. They knew—from genetics—a lot about the competences their quarry had to possess, and they set their task as the construction of a physically possible model of something that would have those competences.

We face a similar task today. Experimental psychology is giving us an ever-more detailed catalogue of the competences and frailties of the mind—the triumphs of perception and the embarrassments of illusion, the pace of language learning and the conditions of distraction, lust, fear, and mirth—and now, as “convinced materialists,” we need to figure out how on earth the brain does it all, without postulating wonder tissue.

As our understanding grows, what counts as wonder tissue shifts. When “connectionist” and other “neural network” models burst on the scene in the mid-1980s, they demonstrated learning capabilities and pattern-recognition powers that nobody would have dared postulate in small tracts of neurons a few years earlier. We still don’t know exactly how—or even if—the brain exploits the computational powers exhibited by these semi-realistic models, but it is now okay to postulate a connectionist competence for some neural network that you can’t yet explain as long as you are up-front about it and the competence is not clearly beyond the demonstrated range of feats. (Randi may not do the trick the very same way Geller does, but we are safe in concluding that there is some variation of Randi’s method that explains Geller’s ability, which gives us guidance in inquiring further into the processes that are actually involved.) The main objection to wonder tissue is that it does not give us a way of solving the problem, but a way of giving up, of assuming that it is a mystery that can never be solved.