EXP: Fostering Self-Correcting Reasoning with Reflection Systems


PI: Michael Hoffmann, Richard Catrambone, Jeremy Lingle
Georgia Tech
Award Details

This research project is exploring how to support reasoning about wicked problems. These are societal important problems that are characterized by incomplete or contradictory knowledge, have a large body of differing opinion on the problem, have a large economic burden, and are intimately interconnected with other problems. An example of such a problem is poverty. Poverty is linked with education, nutrition to poverty, the economy with nutrition, etc. Reasoning about such problems and coming up with partial solutions is an important learning activity. One aspect of approaching wicked problems is through the use of reflection to guide argumentation. This project explores supporting reflection in undergraduate students with software that supports the reflection process and software that aims to improve the quality of arguments. This software builds upon both visualizations of arguments and a structured format, known as the Vee diagram, that structures good argumentation through a process of studying, questions synthesis, and finally analysis and reflection.

More specifically, the researchers analyze how experts approach wicked problems, how they engage in reflection, and how they assess and improve the quality of their arguments. Results of these experiments with experts will be incorporated into Computer Supported Argument Visualization (CSAV) tools. The approaches explored in this project are of two types: the use of templates to trigger reflection and the use of scripts to provide a structure to reason about an issue. As a starting point, the researchers build upon the argumentation Vee diagram for the first approach and the AGORA software, which has been developed by the PI, as an approach for a script-based approach. Results of the experiments will contribute to an understanding of how reflective learning and self-correcting reasoning can be fostered by assessing specific features of reflection tools and interactions scripts. Research results enable known obstacles to self-improvement, such as students’ implicit assumptions about the nature and certainty of knowledge and bias, to be addressed through educational interventions.

Tags: , ,