MODELS OF MY LIFE By Herbert A. Simon. Illustrated. 415 pp. New York: Basic Books. $26.95.
As much as any one person, Herbert A. Simon has shaped the intellectual agenda of the human and social sciences in the second half of the 20th century. A brilliant polymath, Mr. Simon is a Nobel laureate in economics, a pioneer in political science, psychology and management and a founding father of artificial intelligence, the enterprise of trying to build machines that think. Now he has turned to autobiography and thus provides a window onto the subjective side of scientific achievement.

"Models of My Life," one of the books in the Alfred P. Sloan Foundation series, offers dramatic examples of science as the product of the whole scientist, of the fit among personality, scientific problems and strategies for their solution. As the work comes out of the life, so the life is shaped by the work. And while readers may come to this book thinking that artificial intelligence is an impersonal product of technological progress, they will come away seeing that it is a deeply personal enterprise -- an expression of the esthetic of the people who work in it and, more than that, a science of self-reflection, a way of thinking about one's own thought.

Herbert Simon was born into a middle-class German immigrant family in Milwaukee in 1916. From his earliest years, and notably from the days of his first algebra course in high school, he had a strong love of order. He was, for instance, "troubled by the fact that some quadratic equations had two solutions, some had one, and some none." Referring to his young self in the third person as "the boy," Mr. Simon comments that the intensity of the boy's feelings "seemed to reveal something of a Platonist within him, a desire to find pattern, and preferably simple pattern, in the world around him." He recalls that during his undergraduate years at the University of Chicago, in the early 1930's, a period of intense intellectual and political ferment, he argued philosophy with friends who were loyal to an unruly "Aristotelian-Thomist-Catholic-Trotskyism." But his own tastes led him to less esoteric intellectual seas. A political scientist by training, Mr. Simon studied decision making within organizations and used his findings to lay the foundation for a critique of neoclassical economic theory.

Mr. Simon developed a vision of rationalism adapted to 20th-century sensibilities. Neoclassical theory, he believed, had turned such abstract goals as what economists call "optimization" and "maximization" into fundamental laws of human behavior, and it had foundered, as he thought any inquiry into ultimate meaning would, on a preoccupation with absolutes. In his view, people who behave rationally are not optimizing anything at all; they are simply making decisions based on what their environment tells them they can and cannot do. People don't strive for the best, he maintained, they look for what is possible within the bounds of their given situation. Instead of satisfying desires, said Mr. Simon, human beings "satisfice," they search "for 'good enough' actions rather than optimal ones."

Mr. Simon tells us that this idea of "bounded rationality" has been his "lodestar for nearly fifty years." As he sees it, people torment themselves when they ask too-big questions about their ultimate desires. He believes the proper procedure is to determine the boundaries in which rationality can most constructively operate and then to keep one's sights within them. Such theoretical beliefs allow Mr. Simon to have a relatively untroubled eye. "Searching for the best can only dissipate scarce cognitive resources," he writes. Borrowing Voltaire's aphorism, he says that in science as in life, "the best is enemy of the good."

In the mid-1950's, Mr. Simon married his vision of human behavior to the possibilities offered by computer technology. The computer gave powerful expression to Mr. Simon's already-formed intellectual esthetic. His idea that humans solve problems under constraint was easily adapted to the task of programming a computer to solve problems using heuristic search -- a procedure that doesn't involve absolute rules but rules of thumb based on the knowledge of specific and bounded environments. And his idea that observation must be the foundation for abstraction led him to create a research strategy for artificial intelligence that he called "thinking aloud."

In "thinking aloud," Mr. Simon and his colleagues asked people to express their thoughts as they solved logical problems, and then they analyzed the transcripts of these sessions. The protocols of human beings playing games and solving puzzles were used to evaluate computer programs made to follow in human footsteps. Whereas some artificial intelligence researchers were content to leave aside the larger question of whether machines that perform intelligent tasks are doing them in the way people might, Mr. Simon took a very different tack. His artificial intelligence was psychology, not engineering. From the first, he says, "we were interested in simulating human problem solving, and not simply in demonstrating how computers could solve hard problems." His school of artificial intelligence was committed to the view that human intelligence is the manipulation of symbols and the implementation of rules. "For my money," he writes, "to show that something whose behavior looks very complex and erratic is really built from the combinatorics of very simple components is beautiful, not demeaning."

What Mr. Simon saw as self-evident, others found controversial. His critics questioned whether he had really demonstrated that people solve problems the way his programs did, and challenged the premise that a rule-based theory could explain the complexity of human behavior. In artificial intelligence, technical and philosophical arguments commingle. So, in addition to contributing to artificial intelligence as a scientific discipline, Mr. Simon became a chief spokesman for a view of the field as a revolutionary and salutary break with past limitations in philosophy and psychology. In Mr. Simon's opinion, the computer offered the 20th century a solution to the mind-body problem -- "how a physical system can have thoughts." He believed that if one put aside the too-literal analogies between the neurological organization of the brain and the wiring of the computer, there was a "more fruitful analogy": people and computers could be shown to process information and manipulate symbols in the same way.

In December 1955, Mr. Simon and his colleague, the artificial intelligence scientist Allen Newell, succeeded in writing a computer program that could prove mathematical theorems. Mr. Simon celebrated the achievement by walking into class in January and announcing to his students, "Over the Christmas holiday, Al Newell and I invented a thinking machine."

To the rest of the world, he was no less enthusiastic or unequivocal. In a letter to Bertrand Russell written the following October, Mr. Simon stressed that when the thinking machine had proved theorems from Russell and Alfred North Whitehead's "Principia Mathematica," the solutions had involved "genuine 'discovery.' " (To this missive Russell replied: "Thank you. . . . I am delighted to know that 'Principia Mathematica' can now be done by machinery. I wish Whitehead and I had known of this possibility before we both wasted ten years doing it by hand.") In a much-cited 1957 paper, Mr. Simon predicted that in 10 years "a digital computer will be the world's chess champion, unless the rules bar it from competition," and that after chess the possibilities were endless. Within the "visible future," Mr. Simon declared, "machines that think, that learn, and that create" will be able to handle the range of problems "coextensive with the range to which the human mind has been applied."

Things have not worked out as straightforwardly as Mr. Simon expected. A digital computer is not yet the world's chess champion, and it has turned out that it is easier to teach computers to solve logical problems than to recognize everyday objects, easier to make them play chess than make mud pies.

These disappointments have led some critics to claim that there are absolute limits on what computers can do and other critics to say that no matter what computers accomplish they will only be simulating thought. The field has sparked an impassioned and continuing debate about the qualities that might make people different from "thinking machines" in fundamental and unbridgeable ways.
In "Models of My Life," however, Mr. Simon barely acknowledges the four decades of controversy in which he and artificial intelligence have been engaged. He interprets and dismisses most of it as romantic resistance. "I think those who object to my characterizing man as simple want somehow to retain a deep mystery at his core," he writes. "In arguing that machines think, we are in the same fix as Darwin when he argued that man shares common ancestors with monkeys, or Galileo when he argued that the Earth spins on its axis."

Mr. Simon does not make his large claims and predictions in a spirit of hyperbole; in his eyes, modeling elements of human logical capacities in machines is the same as creating intelligence. His simplification of the scientific path ahead and his impatience with skeptics have always been in the service of starting and protecting a new discipline. He sees himself as a pioneer, inventor and missionary, whose job is to present artificial intelligence in as sharp and clear a form as possible. The task of recognizing limitations and pronouncing caveats is better left to others. His role was to announce a scientific revolution and declare it ready to set up provisional governments in all the human and social sciences.

One can hardly imagine a better fit between a man and his intellectual enterprise. Artificial intelligence is the expression of Herbert Simon the economist and Herbert Simon the man. He used thinking about his own thinking and strong personal convictions about bounded rationality to create the general shape of his theory. Conversely, when he contemplates his life he uses concepts drawn from his vision of the field. In "Models of My Life," artificial intelligence appears not only as a technical discipline but also, like psychoanalysis, as a structure for self-reflection.

Through its lens, Mr. Simon sees his life in terms of choices, branch points in a maze, searches and decisions. But there are no demons. Mr. Simon describes his life as a "maze without a Minotaur." Indeed, in 1970, he sought out Jorge Luis Borges in Argentina to discuss their shared metaphor of the labyrinth. For Mr. Simon, the search for patterns and rules goes beyond intellectual matters; he looks to it to provide him with a moral metric and a set of values for living. In his autobiography, he explores the meaning of intimate experiences through the formal propositions that might hold them together. Thus, he seems disappointed that Borges denied there was an abstract model underlying his mazes.

In Mr. Simon's hands, even the most emotionally charged events become occasions to discover rules and defend theorems. For instance, he describes an episode during which he contemplated infidelity with an attractive former student he calls Karen. He begins by saying that he "cannot deny" the proposition that one might be "genuinely in love with two women at once" and then shows how he used his experience to add an important new corollary to his theory of love: "You can love two or more women at once . . . but you cannot be loyal to more than one." In his account, Mr. Simon casts his wife in the role of dialogue partner for himself and Karen as they all agree that he was "asking for the moon," attempting the logically impossible. The consistency of Mr. Simon's scientific and personal esthetic, the penchant for the propositional, will surely draw fire from those with Freudian sensibilities who will be alienated by his search for simple rules in complicated situations and his tendency to close down potentially discomforting questions.

When, in 1965, Mr. Simon visited the school in Darmstadt, Germany, where his father had studied engineering, he was faced with questions about his father's real reasons for immigrating to America. Before leaving Germany, in 1903, his father had challenged a fellow student to a duel because of what Mr. Simon believes was an anti-Semitic comment. What, asks Mr. Simon, was his father's relationship to his Jewishness? What, wonders the reader, is Mr. Simon's own relationship to these elements of his past? But such questions are dismissed as irrelevant. They will not yield up facts: "Nothing we could now seek out in Darmstadt would reveal the secret of that student quarrel." And the feelings they would bring up would only be unhelpful: "The mists perform an important social function when they place a statute of limitations on man's memory of ancient wrongs."

The information-processing lens, like all theories of mind, is selective in what it places in the foreground and leaves in the background. It consistently leads Mr. Simon to see facts as more important than feelings and to see logic taking precedence over all. His lodestar principle of bounded rationality makes him comfortable with a situation in which there is little room for considering those things an information-processing model leaves out. He is comfortable with the mists that hide the Minotaur. In focusing on the maze, he falls squarely on one side of a division in the long history of thinking about human nature -- on the side that sees people as rational animals and rationalism as the source of a happy life. If deviation from rationalism is what gets us into trouble, let's be rational. If memory is painful, let's forget.

Of course, there has always been another approach -- to put oneself in touch with memory and Minotaurs. In this view, dreams of rational progress founder on the shoals of fantasy, passion and irrationality. In choosing a life as well as a science of simple propositions, patterns and principles, Mr. Simon is certainly a pre-Freudian, one who does not recognize a seamless web between rational and irrational. But the more important issue raised by this book is that Mr. Simon and his computational esthetic may be the harbingers of a post-Freudian world view. For many readers, Mr. Simon's view of human endeavor, of love and of work, will seem emblematic not of the pre-Freudian rationalism-that- was but of a new, sleeker rationalism-to-be -- a rationalism purged of utopian excess, committed to empirical studies and wedded to the most modern technology.

"Models of My Life" challenges us to think carefully about the costs as well as the benefits of this new rationalism, by suggesting that it may blind us to the Minotaurs in our mazes. And since the parents of scientific revolutions are often more effective if they gaze on their progeny with untroubled eyes, Mr. Simon's book challenges us to do what he has not.
If there is one thing holding together Herbert A. Simon's work in economics, artificial intelligence and psychology, it is the idea that making decisions is like going through a maze. But even he acknowledged that the maze metaphor can be confining.

After all, "one doesn't spend most of one's time making decisions. One lives. I don't feel like I'm in a maze," he said during a telephone interview from the University of Minnesota in Minneapolis, where he was attending a conference on cognitive psychology. "The choices we make lead up to actual experiences. It is one thing to decide to climb a mountain. It is quite another to be on top of it."
Mr. Simon said that "any model of human behavior that focuses on decision making gives us an overrational idea of humans. Artificial intelligence is only a model of human choice, not of human emotion." So, while a computer program can simulate what human beings are doing when they're making decisions, "no one has ever tried to model, say, two people married to each other."
Why focus on the rational rather than on the emotional? That choice, Mr. Simon said, "may be a commentary on my personality. Being reasonable, if not rational, is very important to me."
His decision to write his autobiography, "Models of My Life," was one of his eminently reasonable decisions. Mr. Simon, a professor of computer science and psychology at Carnegie-Mellon University, said he began to reminisce about his life when Pamela McCorduck interviewed him for her book "Machines Who Think." As he recalled, "she said she wanted to do a biography of me, but then she found something more interesting to do." So Mr. Simon wrote it himself, thinking, as he does about many things in his life, that it was for the best. After all, he said, "I have access to some sources that no one else has."

Mr. Simon said that the experience of writing an autobiography hasn't changed his thinking about thinking and that history hasn't changed his thinking about thinking computers. Though his prediction that a computer would be the world's chess champion by 1967 did not come true, he said, "I still feel good about my prediction. Only the time frame was a bit short." -- SARAH BOXER
Photo: Herbert A. Simon (Basic Books) (pg. 29)