Understanding Explanatory Coherence

During the past five years, Prof. Ranney's interdisciplinary Reasoning Group has primarily focused on studying, modeling, and promoting the coherence of explanations with respect to adolescent and adult learners. Studying explanatory coherence has involved psychological and linguistic analyses regarding the nature of (and relationships between) evidence and hypotheses, among other concepts (such as fact, theory, proof, etc.). Modeling explanatory coherence has been a multidisciplinary effort among philosophers, psychologists, educators, natural scientists and computer scientists. Predictions from an artificially intelligent connectionist computer model (ECHO) have been used to simulate human reasoning about a host of important scientific and social controversies. Combining these preceding efforts, a useful "reasoner's workbench, "Convince Me," has been developed to foster more coherent reasoning in people facing difficult decisions and has been assessed over a wide variety of domains (e.g., reasoning about physics, neurophysiology, geography, the environment, the death penalty, etc.). It has proved highly effective in improving scientific reasoning in users. (Professor Ranney)

Professor Ranney from the EMST (Education in Mathematics, Science, and Technology) division of the School of Education has been studying the kinds of reasoning people use in assessing the coherence of scientific theories and explanations. In collaboration with philosopher Paul Thagard he developed and implemented a connectionist network theory of explanatory coherence called ECHO (Explanatory Coherence by Harmany Optimization). ECHO has been applied in many interesting ways. For instance, it has been used to model students' changing beliefs about the abortion controversy, jury trials, and physical systems. Ranney's group has recently extended this work in their "reasoner's workbench" software (Convince Me) which enables them both to study the coherence of students' reasoning abilities, and to foster it in learning situations. This work includes elements of computational modelling, psychology, philosophy, linguistics, and education.