The premise of David Gelernter's The Muse in the Machine is that artificial intelligence research is incomplete, in that it fails to address traditionally non-scientific forms of human thought like emotion, metaphoric imagination and hallucination.
Human thought, asserts Gelernter, exists along a continuum, spanning from high-focus thinking -- in-depth analytical problem solving -- to low-focus thoughts, consisting of the daydreams and hallucinations that occur when one's mind is wandering. Artificial intelligence research has historically focused on the logical, goal-driven thoughts at the high-end of the spectrum. Gelernter argues that, if the goal truly is to create a computer that can solve problems in a way similar to that of the human mind, then study of unfocused, emotional, low-end thinking must be incorporated into artificial intelligence research, for it is as central to human cognition as logic.
Gelernter compares his thought spectrum to the types of thinking that people do daily. High-focused thought is employed when someone is alert and focusing on a problem. Low-focus thinking takes over when we are tired or falling asleep and appear to have no control over the pattern our thoughts take. It is at the latter point that we are at our most creative. Gelernter defends his assertion in a gripping passage (p.24-26) where he cites the ideas of romantic poets on the role of emotion in human thought. He draws further analogies between his thought spectrum and human cognitive development. Children are more creative than analytical, but develop logical problem-solving skills as they grow to adulthood. (p. 13) His third, and most debatable, analogy likens the intellectual development of mankind as a journey from the dream-like thinking apparent in some passages of the Bible, what Gelernter calls, "prelogical antiquity" (p.15) to the logical thought valued today.
The Muse in the Machine was intended for a nonscientific audience. "Virtually everybody is interested in the mind, has wondered about their own thought processes, but the problem is the topic has been hijacked by scientists," Gelernter told the New York Times. (Weaver, Jacqueline, Connecticut Q & A: David Gelernter; NYT, 7/31/94) The author succeeds in composing a beautifully written work that could be of interest to anyone with curiosity about artificial intelligence and the human mind, but the book should not be viewed as an introduction to the topic. Muse is highly critical of the field of AI, and does not attempt to offer balanced scholarly discussion of the criticisms it asserts. The book is visionary and thought-provoking, however, and a pleasure to read.
David Gelernter is a professor of Computer Science at Yale University and is the developer of Linda, a computer language designed to help facilitate communication between computers for parallel processing.