Responding to Student Affect and Efficacy Through Empathetic Companion Agents in Interactive Learning Environments

Abstract

Because many students experience frustration during learning, it is important to develop affective strategies to support students ’ coping with frustration in interactive learning environments. First, we must devise affect recognition models to detect student affect. Second, we need to determine when to intervene; these conditions are likely to be different for each student. To determine how much frustration a student can persist through, we should utilize models of student self-efficacy to predict a student’s frustration threshold. Third, we should devise techniques for responding empathetically before the student reaches her threshold of frustration. We propose an approach to support students ’ coping with frustration in intelligent tutoring systems that utilizes induced models of affect, self-efficacy and empathetic behavior to effectively reason about precisely when and how to intervene in frustration-ridden learning situations.