When robots tell you what to do: Sense of agency in human- and robot-guided actions

The present study investigated the sense of agency (SoA) when actions were determined by another human vs. a humanoid robot as compared to when freely selected. Additionally, perceived robot-autonomy was manipulated via autonomous vs. non-autonomous descriptions of the robot. SoA was assessed by judgment of control ratings and intentional binding (i.e., perceived temporal attraction between voluntary actions and their outcomes). Participants performed free and instructed key presses that produced an auditory tone (Experiment-1) and visual stimuli conveying neutral, positive, or negative valence (Experiment-2). Binding and control ratings were greater in free compared to instructed actions, and comparable between human- and robot-instructed actions. Control ratings were higher for positive compared to neutral and negative outcomes, and positively correlated with ratings of how human-like the robot appeared. These results highlight the role of endogenous processing of action selection and provide preliminary insight into the SoA when actions are guided by artificial agents.