Abstract
This study explored the temporal dynamics of motor imagery (MI) following the observation of robotic versus human actions using a temporal judgment task. Participants observed goal-directed action videos by a human or a NAO robot, performed MI until an auditory stop signal, then completed a two-alternative forced choice task to identify the frame corresponding to their imagery stop point, choosing between the correct frame and either a preceding ('Before') or succeeding ('After') frame. Results revealed a systematic temporal bias in MI towards 'After' incorrect frames. Crucially, this bias was significantly smaller for imagining robotic actions compared to human actions. This difference was primarily driven by an increased number of 'Before' errors for longer robotic actions, suggesting a perceived slowing down of MI (leading to an attenuated forward bias) when simulating less biologically plausible robotic movements, especially over extended durations. For human actions, video duration did not significantly modulate temporal errors. This finding indicates that the temporal distortion in MI is modulated by the observed agent's movement characteristics and action duration, challenging simplistic embodied simulation accounts.