We believe that in order to analyze creative reasoning, one needs a theoretical computational framework in which to model thinking. To this end, we propose using a computational approach rooted in case-based reasoning. This paradigm is fundamentally concerned with memory issues, such as remindings from partial matches at varying levels of representation and the formation of analogical maps between seemingly disparate situations -- exactly the kinds of phenomena that are central to creativity.
Accordingly, we see creative thought, like all thought, as involving processes of problem interpretation and problem reformulation, case and model retrieval, elaboration and adaptation, and ultimately, evaluation. Research in case-based reasoning has provided extensive knowledge of how to analyze and reformulate problems, how to reuse solutions to old problems in new situations, how to build and search libraries of experience, how to merge and adapt experiences, and how to evaluate candidate solutions. We are taking primarily a case-based approach to modeling creativity, complemented by research in model-based reasoning, meta-cognition, visual reasoning, and thought experimentation.
We are studying creativity in three disparate domains: We are studying creativity in the everyday activities of average people by studying the design of mechanical devices and by looking at the processes involved in reading and understanding science fiction stories. At the same time, we are examining and analyzing what led to the significant scientific discoveries of Maxwell and Faraday.