Enhancing Learner Models for Pedagogical Agent Scaffolding of Self-Regulated Learning

Abstract

Self-regulated learning (SRL) processes for monitoring and modulating one’s own cognition, affect, metacognition, and motivation are essential for effective learning within intelligent tutoring systems (ITSs). As such, pedagogical agents are typically embedded within these environments to scaffold learners’ SRL via prompts to engage in these processes. However, current pedagogical agents follow pre-established, production rules based on temporal and frequency inputs from the learner which limits the ability for these agents to provide individualized scaffolding intelligently and adaptively. In this paper, we argue that this issue originates from an underdeveloped learner model that does not comprehensively track the dynamic nature of SRL. This paper introduces recurrence plots, which visualize system dynamics, as a novel method for enhancing learner models to provide pedagogical agents the information necessary to interpret the dynamics of learners’ enacted SRL processes captured using log files as they engaged with instructional materials during learning with an ITS. In merging recurrence plots and the field of SRL, learner models can be enhanced with information regarding learners’ dynamical use of SRL processes to augment the accuracy and sophistication of pedagogical agent scaffolding.