Intelligence is a function of stories, states the author (director, Institute for the Learning Sciences, Northwestern Univ.)--how many we know, how efficiently we compare them with new ones, and how we modify our stories as a result. People, Schank explains, constantly assess others' intelligence through the ways in which they respond in conversation. Appropriate, predictable answers might signify moderate intelligence. Unusual, creative connections that lead to unexpected responses give an impression of greater intelligence. In fact, these responses--the voicing of common elements between others' stories and our own--better gauge people's true intelligence, claims Schank, than the standard tests of verbal and spatial abilities used in schools. We instinctively tell stories to define ourselves and others, to influence our listeners, to summarize events, to control a conversation, etc. Our listener's only way of understanding our stories is to match aspects of them with aspects of the ones the listener already knows. Failure to do so signals ignorance (the listener hasn't learned enough stories) or lack of intelligence (the listener is unable to efficiently index elements of the stories he or she has). A mechanistic model, this, but surprisingly persuasive--particularly when Schank extends his theorizing to propose as teaching tools interactive storytelling computers that can respond to their users' cues, telling different stories to different people in ways that those people can best understand. "Human beings are naturally predisposed to hear, remember, and to tell stories. The problem for teachers, parents, government leaders, friends, and computers is to have more interesting stories to tell." Certainly this one is intriguing.