Frontal and occipito-temporal Theta activity as marker of error monitoring in Human-Avatar joint performance

Discrepancies between sensory predictions and action outcome are at the base of error coding. However, these phenomena have mainly been studied focusing on individual performance. Here, we explored prediction errors during a human-avatar motor interaction and focused on both the classical frontal error-related brain responses and the activity of the action observation network. Our motor interaction paradigm required healthy individuals to synchronize their reach-to-grasp movements with those of a virtual partner in conditions that did (Interactive) or did not require (Cued) movement prediction and adaptation to the partners actions. Crucially, in 30% of the trials the virtual partner suddenly and unpredictably changed its movement trajectory thereby violating the human participants expectation. These changes elicited error-related neuromarkers (ERN/Pe - Theta/Alpha modulations) over fronto-central electrodes mainly during the Interactive condition. Source localization and connectivity analyses showed that the frontal Theta activity induced by violations of the expected interactive movements was in phase with occipito-temporal Theta activity. These results expand current knowledge about the neural correlates of on-line motor interactions linking the frontal error-monitoring system to visual, body motion-related, responses.