The Brain's Software

October 5, 1997
By Mark Ridley
 

HOW THE MIND WORKS 
By Steven Pinker.
Illustrated. 660 pp. New York:
W. W. Norton & Company. $29.95.
 

"Interdisciplinary" is a suspect word -- it is too often the grayspeak of campus grandees -- but artificial intelligence and the theory of evolution do seem to be making an interdisciplinary merger. Steven Pinker, a professor of psychology at the Massachusetts Institute of Technology, is one of the clearest writers about this new synthesis. "How the Mind Works" examines brain mechanisms -- or rather, computational models of brain mechanisms -- and the evolution of our reasoning abilities, emotions and social and sexual behavior. The marriage is by no means complete; Pinker exaggerates when he says that "without the computational theory, it is impossible to make sense of the evolution of the mind." I suspect the evolutionary insights in the book would have made good sense to that precomputational Victorian Charles Darwin. However, there is still a strong unity in this book. The computational and evolutionary parts, for instance, both reject the idea that the mind begins life, in Locke's phrase, as a sheet of blank paper.

Artificial intelligence is a powerful test; it requires you not merely to talk about how a mental task is accomplished, but to write a program to prove it. If you program a robot as initially blank except for a general learning rule or two (I mean something like the psychologist's "law of association": the robot should do more of whatever it has associated with pleasurable, or reinforcing, consequences), it will not develop much of a mind. It fails because the environment contains too many stimuli, and each action has too many possible consequences. Unless the robot has been programmed to recognize which stimuli to concentrate on and how to assess them, it is hopeless at human tasks.

Indeed, for many tasks, including walking or holding or reading a newspaper, the robot will be hopeless even if it contains state-of-the-art software. Humans find them easy enough in practice, but the rocket scientists are baffled about how we manage them in theory. As for walking, "no one has yet figured out how we do it." When you hold up your arm, your brain is "solving a near-intractable physics problem." When you use your eyes, you solve "what engineers call an 'ill-posed problem.' It literally has no solution." Pinker adds, "I believe that the discovery by cognitive science and artificial intelligence of the technical challenges overcome by our mundane mental activity is one of the great revelations of science, an awakening of the imagination comparable to learning that the universe is made up of billions of galaxies or that a drop of pond water teems with microscopic life." He concludes that the mind must possess a set of specialized, preprogrammed modules that are quite unlike one another, each designed for a task such as perception or social exchange.

Evolutionary theorists reject the blank-paper theory for a related reason. Behavior is adaptive -- well designed to enable the animal to live and breed. Adaptive behavior is not any old behavior; most of the things an animal could do would make it more likely to die. If animals began life with white-paper minds, there would be no reason for them to learn adaptive rather than unadaptive behavior; they must have, in Konrad Lorenz's phrase, an "innate schoolmarm" to teach them what to learn.

But is human behavior adaptive in the way that Darwinian theory predicts? Many critics doubt it. Some, like Alfred Russel Wallace (co-discoverer of natural selection with Charles Darwin), suggested that our brains are more powerful than is needed for survival and reproduction, and that we have more elevated tastes than we absolutely need. Why should natural selection supply us with the ability to acquire appreciation for art or a skill in calculus? Others have made the opposite case: we are too irrational, too ready to believe in absurd animistic causes and too dumb when it comes to probability and statistics.

I am persuaded by Pinker's main claims, that our minds evolved by natural selection and that our mental abilities require elaborately designed programs. However, I have some fairly high-level disagreements. I could pick an argument with his theory of the emotions, with his idea that our mental adaptations are out of date in modern Western society, with his evidence from questionnaires, with his generally dismissive attitude toward cultural influence. But I'll pick on another issue: modifiability. Pinker includes in a list of the idiotic beliefs held by other scholars "that people could just as easily be conditioned to enjoy the thought of their spouse being unfaithful as to be upset by the thought." He discusses how natural selection will theoretically favor sexual jealousy, and how the facts (Margaret Mead and disciples notwithstanding) match the theory. Sexual jealousy, I agree, is a Darwinian adaptation that enabled some ancestral humans to outreproduce their more relaxed contemporaries, who did not end up among our ancestors.

But to show something is an adaptation says nothing about how easy it may be to modify. In evolutionary theory, people learned sexual jealousy in the range of environments occupied by ancestral humans. But what about the new environmental possibilities of the present? Consider, as an analogy, the genetic disease phenylketonuria, which renders people unable to metabolize phenylalanine, with potentially disastrous results. Normal diets contain phenylalanine, and for those diets it would be idiotic to say that people with the gene for phenylketonuria could just as easily grow up with or without phenylketonuria. Once the disease was understood, diets were engineered that lacked phenylalanine; the disease can be almost eliminated by these diets, and it is now in a sense equally easy for someone with the gene to grow up with or without phenylketonuria. Jealousy certainly develops in all the social environments we know. But technology can create new environments. If we understood how jealousy develops, we might be able to modify it. I can imagine a society in which people are conditioned to enjoy the thought of their spouse's being unfaithful.

Elsewhere in the book Pinker is admirably clear on a related point, that to show that natural selection favors something is not the same as to justify it morally. He describes how he personally has ignored "the solemn imperative to spread my genes. By Darwinian standards I am a horrible mistake. . . . But I am happy to be that way, and if my genes don't like it, they can go jump in the lake."

READERS of Pinker's earlier book "The Language Instinct" will be delighted to see there is "a new Pinker." After that masterpiece, expectations are inevitably high, but I was not disappointed. "How the Mind Works" is just as literate -- witty popular science that you enjoy reading for the writing as well as for the science. Pinker has breathed marvelous life into the computational models, the originals of which are buried in nerdish obscurity. He knows when to hold his readers' attention with an illustration or a joke. No other science writer makes me laugh so much. He has good movie quotes -- whether a Mae Westism ("Men like women with a past because they hope history will repeat itself") or the theory of why handsome men pay prostitutes, posed in a movie about Heidi Fleiss ("They're not paying you for the sex. They're paying you to go away afterwards"). His politics do not appear to be conservative (indeed, he has some liberating ideas for both feminists and liberals), but he can be delicately politically incorrigible. He refers to "rain forests -- or, as they used to be called, jungles," and to "knickknacks in the shapes of animals and naked women, which archeologists euphemistically call 'fertility symbols.' " He also neatly insinuates that we could equally well call pornography a fertility symbol. He is a top-rate writer, and deserves the superlatives that are lavished on him.
 

                  Copyright 1997 The New York Times Company