The Emotion Machine: Common-sense Thinking, Artificial Intelligence and the Future of the Human Mind
Simon & Schuster / 387 pages / $26
Reality leaves a lot to the imagination," John Lennon once quipped. That may be why the human mind has developed so many different ways of responding to the external world. The brain, according to Marvin Minsky, a professor of media arts and sciences, electrical engineering and computer science at the Massachusetts Institute of Technology, addresses problems by selecting from its tool kit of resources, which includes instincts, memories, analogies and "common sense." Intuitions, feelings and emotions constitute thinking in other forms. Some day, Minsky predicts, experts in artificial intelligence will be able to re-create the whole range of "human" cognition.
In The Emotion Machine Minsky speculates about simple and complex ways of thinking. At times disorganized, the book is informative, ingenious and accessible to a general audience. Like animals, Minsky observes, people exhibit instinctive reactions: They hear a sound and turn toward it. From parents (or other "imprimers") they learn to react as well, looking both ways before they cross the street. Deliberative thinking allows human beings to consider alternatives before making a decision. Reflective thinking permits an evaluation not only of external phenomena but of activities inside the brain. And self-reflection thinking makes it possible to consider a decision in light of a person's self-image. Confusion and conflict, Minsky maintains, force us to use some (cognitive) roads less traveled by. Positive reinforcement, by contrast, can lead to rigidity.
Every person, Minsky indicates, is "a river of rivaling interests." So the brain must manage deliberation and reflection. The brain knows, somehow, that we can walk while we talk but not while we write. "Action Planners" compose a sequence of "Motion Goals," helping us move through a crowded room without bumping into the furniture. At higher levels of cognition, Minsky guesses, the brain retrieves memories just before we need them: Although it "sees" only a collection of pigment spots, "we re-cognize things by being re-minded of familiar objects that would match incomplete fragments of evidence."
Minsky asserts, provocatively and persuasively, that thinking occurs when "Critics" in the brain shut down some resources while turning to others. "Correctors" alert us to danger; "Suppressors" interrupt actions: and "Censors" prevent ideas from occurring to us in the first place. Logic, Minsky points out, involves a chain of reasoning that is only as strong as its weakest link. And there are exceptions to every rule. "So, using logic is somewhat like walking the plank; it assumes that each separate step is correct." And it relies on interventions by the brain. If too many "Critics" weigh in, the brain describes the problem in more detail; if too few "Critics" are aroused, the brain generalizes.
Emotions also depend on limiting some brain activities. Marathoners and mountain-climbers experience pleasure by suppressing pain. Meditators turn their "Critics" off. As do lovers and those who are deemed "decisive." They stop examining alternatives, "irrelevant" details, and other information that might challenge the status quo.
"Common sense," Minsky writes, is perhaps the most sophisticated form of knowledge. It interprets every piece of information in the context in which it is encoded and decoded. He uses an excerpt from a children's book to illustrate: "Mary was invited to Jack's party. She wondered if he would like a kite. She went and shook her piggy bank. It made no sound." A reader with common sense "knows" that Mary wanted to give Jack a present; she went to her piggy bank for money, not because it contained a kite; the bank would have rattled if it contained coins; and Mary now realizes she does not have the money to buy the kite.
Although he can define common sense, Minsky cannot explain how it works. Or the mechanics of storing memories and retrieving them. To make fragments of knowledge readily available, he hypothesizes, the brain may attach them to goals, situations in which they have appeared, costs, side effects, and common exceptions. Minsky also does not know how the brain makes, organizes and changes its ways of thinking. Or whether some parts of the brain recognize when others are performing poorly.
Neurobiologists learn more about the brain every week. But, Minsky writes, "they do not yet know enough to simulate even a spider or a snake." Nor can experts on artificial intelligence build a "baby machine" that learns by interacting with the real world the way an infant does. But Minsky is undaunted. He recommends that designers abandon the notion of a "coherent essence" that holds brains together, and design a machine with semi-autonomous parts and pathways, checks and balances, redundancies and default options. We won't understand human brains, he concludes, unless we keep trying to build them. Every system "will surprise us with new kinds of flaws - until those machines become clever enough to conceal their faults from us." It's a heady - and scary - thought.
Glenn C. Altschuler is the Thomas and Dorothy Litwin Professor of American Studies at Cornell University.