Supporting math problem solving coaching for young students: A case for weak learning companion

Abstract

Solving challenging non-routine math problems often invites students to ride an “emotional roller coaster” to experience rich sets of emotions including confusion, frustration, surprise and joy. If done right, it stimulates young students’ curiosity and interest in math and cultivate perseverance and resilience with long-term impact. Effective coaching needs to resolve an instance of “assistance dilemma [1]: making real time decisions on the right type of supports, be it cognitive, meta-cognitive or social, at the right time in order to maximize students’ exposure to “productive struggles” while minimize unproductive ones. Though this ideal model of coaching is possible one-on-one basis, it is often not realistic in a regular classroom with high student-to-teacher ratio. In this thesis, I plan to explore a weak form of learning companion that can actively monitor students behavior to assist teacher to decide who to help and provides non-cognitive supports when teacher is not available.