Predicting beyond ourselves: How expectations shape the perception of external action-outcomes

Abstract

Goal-directed actions are an essential part of our everyday life. They free us from enslavement to sensory input and instead allow us to pro-actively influence and shape the world around us. Whenever we act, we generally also know and feel that we are in control of our actions and the following consequences. This pervasive experience is known as the sense of agency, and has an important impact on many facets of our lives, including our mental health and our justice system. These broad implications raise the question how the sense of agency emerges. In this dissertation, I examined the potential sources of this experience, and particularly focused on the dominant view that agency experiences result from predictions generated by our motor system. According to this account, (life-long) exposure to actions and following consequences allows the brain to form associations between these events. Based on these associations the brain can subsequently already predict an upcoming sensory action-effect from the motor commands that precede the action. Whenever there is a match between expected and actual outcomes, we tend to ascribe outcomes to ourselves (i.e., experience agency). I evaluated the influence of motor predictions on the sense of agency by means of implicit measures of this experience. Instead of relying on self-report, such implicit measures assess the perceptual processing of action-outcomes, which is then used as a proxy for experienced agency. Across multiple studies, and in various sensory domains, I observed only relatively weak evidence for the influence of motor predictions on perception. For instance, participants did not perceive a difference in intensity between tones that were congruent versus incongruent with previously learned action-tone associations. The research conducted and reviewed for this dissertation instead suggests that motor predictions are of less importance for the processing of action-outcomes than is currently acknowledged. This seems particularly true for environment-related action-outcomes (e.g., tones or visual stimuli produced by button presses), which, in contrast to body-related action-outcomes (e.g., self-touch and vocalization), have no clear one-on-one relationship with preceding actions. I propose that while motor predictions might modulate the processing of body-related action-outcomes, the processing of environment-related action-outcomes is more likely to rely on more general, non-motor predictive mechanisms. These insights add to a growing body of work suggesting that the role of motor predictions in (implicit) agency experiences might need to be reconsidered.