#974: A Multimodal Study of Emotion Regulation in Student Dyads During Computational Model Building

Collaborative problem solving (CPS) requires coordination of cognition and affect in complex, open-ended tasks. This study examines how pairs of high school students regulate learner-centered emotions (engagement, confusion, frustration, boredom, and delight) while constructing computational models in a STEM+C context. Extending Gross’s (2015) process model to the group level, we analyze temporal dynamics of synchronized and unsynchronized emotional states during CPS. Multimodal data from six dyads (video, audio, system logs) were coded for CPS activities (e.g., information sharing, consensus building) and emotion transitions. Emotional alignment predominated, yet brief misalignments (especially between engagement and confusion) were frequent and often precipitated interpersonal regulation. Temporal analysis identified two dominant pathways: (1) transitions from misalignment to alignment that restore shared engagement, and (2) transitions between aligned states that shift partners from shared confusion to engagement. These pathways primarily occurred during information pooling and integration phases. Quantitative results show that effective regulation predicts higher model-building performance, with sustained engagement emerging as the stable endpoint of productive collaboration. The findings provide a temporal, group-level account of how students collectively manage emotion to sustain effective problem solving.