A computational model of scientific insight
Scientific discoveries often take the form of insight, in which previously unseen and unexpected connections suddenly reveal themselves to the mind. In this paper, we present a computational theory of this phenomenon. We recount a number of well-known examples of the process, along with some early and recent theories that attempt to explain the phenomenon. However, our reservations about these theories have led us to develop an alternative model. We explain insight as the sudden retrieval of an analogy from long-term memory. To model this retrieval, we use a form of spreading activation. In constructing our framework, we have built on research in two separate lines of research in cognitive science on reasoning by analogy and qualitative models. Thus, we also review some work in these areas.