Modeling one-on-one tutoring sessions

The overarching goal of this project is to develop computational models of the nonverbal behavior and interactive strategies observed during face-to-face teaching. This project will help advance the science of learning and teaching by improving our understanding of the dynamics of nonverbal behavior in teaching at a computational level across multiple scales, including low-level facial movements, cognitive and affective processes, and higher level strategic behaviors. In this work we connect higher level teacher and student behaviors to lower level eye gaze dynamics to inform the development of models of nonverbal behavior. Specifically, we use student and teacher behaviors to predict teacher-to-student gaze onset. We additionally model student and teacher behaviors as a probabilistic finite state machine to examine cross-session teaching policy. Findings suggest a relationship between tutoring events and teacher gaze to student, with cross-validation yielding a mean 2AFC performance of .69. Future analysis will connect automated detection of facial affect to the behavioral events explored in this paper.