Temporal Dynamics of Brain Activity Predicting Illusory Agency over Involuntary Movements

Our muscles are the primary means through which we affect the external world, and the sense of agency (SoA) over the action through those muscles is fundamental to our self-awareness. However, SoA research to date has focused almost exclusively on agency over action outcomes rather than over the musculature itself, as it was believed that SoA over the musculature could not be manipulated directly. Drawing on methods from human-computer interaction and adaptive experimentation, we use human-in-the-loop Bayesian optimization to tune the timing of electrical muscle stimulation so as to robustly elicit an illusion of agency over electrically-actuated muscle movements. We use time-resolved decoding of subjects’ EEG to estimate the time course of neural activity which predicts illusory agency on a trial-by-trial basis. Like paradigms which assess SoA over action consequences, we found that the late (post-conscious) neural activity predicts SoA. Unlike typical paradigms, however, we also find patterns of early (sensorimotor) activity with distinct temporal dynamics predicts agency over muscle movements, suggesting that the “neural correlates of agency” may depend on the level of abstraction (i.e., direct sensorimotor feedback vs. downstream consequences) most relevant to a given agency judgement. Moreover, fractal analysis of the EEG suggests that SoA-contingent dynamics of neural activity may modulate the sensitivity of the motor system to external input. ### Competing Interest Statement The authors have declared no competing interest.