Scientific problem solving by expert systems



Human expert problem solving in science is defined and used to account for scientific discovery. These ideas, attributed largely to Herbert Simon, are used in a description of BACON.5, a machine expert problem solver that discovers scientific laws using data-driven heuristics and “expectations” such as symmetry. Trial-and-error search of data-driven scientific discovery is drastically reduced when the BACON.5 system is altered to include expectations, such as symmetry, which have been influential in “real” (i.e., human) discoveries of scientific laws. A discussion of the implications of BACON.5-type research for traditional science education research recognizes the importance of a qualitative understanding of the relationships among “pieces” of a physical or biological system.