Instructor-learner brain coupling discriminates between instructional approaches and predicts learning

Abstract The neural mechanisms that support naturalistic learning via effective pedagogical approaches remain elusive. Here we use functional near-infrared spectroscopy to measure brain activity from instructor-learner dyads simultaneously during dynamic conceptual learning. We report that brain-to-brain coupling is correlated with learning outcomes, and, crucially, appears to be driven by specific scaffolding behaviors on the part of the instructors (e.g., asking guiding questions or providing hints). Brain-to-brain coupling enhancement is absent when instructors use an explanation approach (e.g., providing definitions or clarifications). Finally, we find that machine-learning techniques are more successful when decoding instructional approaches (scaffolding vs. explanation) from brain-to-brain coupling data than when using a single-brain method. These findings suggest that brain-to-brain coupling as a pedagogically relevant measure tracks the naturalistic instructional process during instructor-learner interaction throughout constructive engagement, but not information clarification.