On Convergence of Cognitive and Non-cognitive Behavior in Collaborative Activity

We present results from a pilot study to investigate the evidence for convergence and synchrony in cognitive and noncognitive behavior of dyads engaged in a collaborative activity. Our approach utilizes multimodal data including video and participant action log files retrieved from the collaborative activity, an online educational simulation on science topics. The log files captured cognitive behavior including frequency and content of chat messages between dyads, as well system help requests. The video data recorded participant nonverbal behavior that was processed on a frame-by-frame basis using automated facial expression classifiers and coded by trained human raters on high-level noncognitive behaviors including: affect display gestures, engagement, anxiety and curiosity. The data were analyzed at individual and dyad levels and results using hierarchical clustering analysis demonstrate evidence of cognitive and noncognitive behavioral convergence among dyads.