The proposed research develops and tests an account of explanation to better understand its role in cognition. The central hypothesis is that explanations have certain properties that serve as a mechanism for the development of knowledge structures that are useful in the sense that they support generalization, prediction, and intervention. The primary question that this research asks is how explanation might contribute to the formation of such knowledge. Explanations are evaluated on the basis of several explanatory virtues - properties that increase the perceived quality of explanations. The proposed research considers two cues: an explanation's simplicity and its breadth or ability to unify diverse phenomena. Both are invoked in science and philosophy of science, and are justified on normative grounds within statistic and computer science.
The study describes three kinds of studies. Lab studies that will help identify features of preferred explanations, a more naturalistic study of explanations that are sought and produced via an online environment, and an experiment to compare conditions in which learners are prompted to generate an explanation or listen to facts. The goal is to understand both the function and content of explanations.
