When temporal prediction errs:ERP responses to delayed action-feedback onset

Sensory suppression effects observed in electroencephalography (EEG) index successful predictions of the type and timing of self-generated sensory feedback. However, it is unclear how precise the timing prediction of sensory feedback is, and how temporal delays between an action and its sensory feedback affect perception. The current study investigated how prediction errors induced by delaying tone onset times affect the processing of sensory feedback in audition. Participants listened to self-generated (via button press) or externally generated tones. Self-generated tones were presented either without or with various delays (50, 100, or 250 ms; in 30% of trials). Comparing listening to externally generated and self-generated tones resulted in action-related P50 amplitude suppression to tones presented immediately or 100 ms after the button press. Subsequent ERP responses became more sensitive to the type of delay. Whereas the comparison of actual and predicted sensory feedback (N1) tolerated temporal uncertainty up to 100 ms, P2 suppression was modulated by delay in a graded manner: suppression decreased with an increase in sensory feedback delay. Self-generated tones occurring 250 ms after the button press additionally elicited an enhanced N2 response. These findings suggest functionally dissociable processes within the forward model that are affected by the timing of sensory feedback to self-action: relative tolerance of temporal delay in the P50 and N1, confirming previous results, but increased sensitivity in the P2. Further, they indicate that temporal prediction errors are treated differently by the auditory system: only delays that occurred after a temporal integration window (∼100 ms) impact the conscious detection of altered sensory feedback.