Exploring Relationships Between Temporal Patterns of Affect and Student Learning

Abstract

Numerous prior articles have studied the relationships between student affect and various outcomes of learning. Prior research has found these relationships are complex; shifts in student affective experiences and the duration in which students remain in particular states form temporal patterns that are often difficult to interpret. Much of the existing research in this area focuses on the correlation between the overall prevalence of particular affective states and learning outcomes, ignoring the temporal and sequential characteristics of student affect. In this work, we leverage temporal clustering methods to identify emerging patterns of affect while students work within a computer-based learning platform to explore how sequence patterns of student concentration, boredom, confusion, and frustration correlate with student performance on a delayed assessment test. Similar to prior work, we find strong relationships between affect and student learning, even accounting for a measure of prior knowledge. Additionally, we identify that the directionality of these correlations in regard to specific affective states differs across clusters. While some affective states such as boredom are identified to exhibit negative relationships with learning within some patterns, a strong positive relationship between frustration and learning is found within one of the emerging pattern clusters.