Dissertation Defense-Quentin King-Shepard

Dissertation Defense By Quentin King-Shepard Friday, July 31st, 2026, at 10 am Murdoch Building, Room 424 Investigating the Role of Metacognition in Collaborative Learning with a Teachable Robot Engaging in metacognitive regulatory skills, planning, monitoring, and evaluation, has been shown to support deeper learning and academic achievement. While much research has examined the individual use of metacognition during learning, one growing body of literature has focused on social metacognition, which involves monitoring and regulating the knowledge and actions of others within collaborative learning environments. Pedagogical agents have been shown to facilitate social behavior during collaborative learning with teachable agents, agents that can take the role of teacher assistant or student co-learner, being shown to enhance students' regulatory skill use. Yet research focused on teachable agents providing training in the use of these regulatory skills has been conducted almost entirely with individual learners and has typically inferred metacognitive engagement from later performance rather than measuring strategy use as it occurs. Consequently, it remains unclear whether metacognitive prompting produces effects that are strategy-specific or strategy-adjacent and whether such prompting supports strategy use while present or scaffolds strategy use that persists once prompting is reduced. In bringing these strands of research together, the present work examined the effects of metacognitive prompting delivered by a teachable robot during collaborative problem-solving activity. Across two experiments, undergraduate dyads taught a NAO robot, named Emma, to solve math problems. During this time, Emma delivered pre-scripted planning, monitoring, and evaluation prompts to student dyads. Interactions were transcribed and coded for both the type of prompt given and the type of metacognitive strategy students engaged in. Zero-inflated negative binomial mixed-