Why learning with others works: interactivity may not be the answer

Social interaction plays a key role in human learning. De Felice et al. [1] showed that learning during a contingent (live) video call outperforms learning from a pre-recorded video of the same session. However, contingent exchanges – when people are part of a live social context – can vary widely in how interactive they are, that is, in the extent to which interlocutors provide mutual, moment-to-moment feedback. In this preregistered study, we performed a post-hoc analysis of De Felice et al. [1]’s dataset and asked whether the learning benefit is driven by the degree of interactivity between learner and teacher. Interactivity was quantified through behavioural annotation of video calls (n = 49) using an annotation scheme developed for this study. The interactivity score reflected a multimodal dyadic measure of teacher-learner verbal grounding (utterances reflecting understanding), non-verbal grounding (nodding and smiling) and questions. Linear mixed-effects models revealed no relationship between interactivity and performance, suggesting that interactivity may not be necessary to drive the social learning advantage. Learning with others may be beneficial because individuals can interact. The potential to interact may prompt cognitive processes specific to live exchanges, such as mutual prediction, that support learning, even if little interaction actually takes place.