Virtual body ownership affects the processing of sensorimotor contingencies and goal-oriented information extraction, but not semantics

Previous research has attempted to quantify presence in virtual environments using event-related potentials (ERPs). We contend, however, that previous efforts have fallen short of fully realizing this goal, failing to either (A) independently manipulate presence, (B) validate their measure of presence against traditional techniques, (C) adequately separate the constructs of presence and attention, and/or (D) implement a realistic and immersive environment and task. We address these shortcomings in an ERP experiment in which participants play an engaging target shooting game in virtual reality (VR). ERPs are time-locked to the release of a ball from a sling. We induce breaks in presence (BIPs) by freezing the ball's release on a minority of trials. Embodiment is manipulated by allowing manual manipulation of the sling with a realistic avatar in one condition (embodied condition) and passive manipulation with only controllers in another (non-embodied condition). We find, in line with our preregistered predictions, that the N2, the P3b, and the N400 are selectively sensitive towards embodied BIPs, information extraction, and BIPs regardless of embodiment, respectively. The pattern of findings carries significant implications for theories of presence, which have been seldom addressed in previous ERP investigations on this topic.