Computational Philosophy of Science

By applying research in artificial intelligence to problems in the philosophy of science, Paul Thagard develops an exciting new approach to the study of scientific reasoning. He uses computational ideas to shed light on how scientific theories are discovered, evaluated, and used in explanations. He describes a detailed computational model of problem solving and discovery that provides a conceptually rich yet rigorous alternative to accounts of scientific knowledge based on formal logic. The model is used to illuminate such topics as the nature of concepts, hypothesis formation, analogy, and theory justification.Following a critique of the alternative account of scientific development offered by evolutionary epistemology, Thagard discusses philosophical issues concerning reasoning, truth, and the justification of scientific methods. He applies his general conclusions about science and pseudoscience to the fields of psychology and artificial intelligence, and explores the potential relevance of computational models to our understanding of the interrelations of theory and experiment and of the importance of group rationality in science."Computational Philosophy of Science" has been made accessible to readers from different disciplines through an appendix that includes tutorials on essential philosophical, computational, and psychological topics.Paul Thagard is a research scientist at the Princeton University Cognitive Science Laboratory. He is coauthor, with John H. Holland, Keith J. Holyoak, and Richard E. Nisbett, of "Induction: Processes of Inference, Learning, and Discovery (MIT Press/Bradford Books). A Bradford Book.