Omission P3 after voluntary action indexes the formation of action-driven prediction

When humans frequently experience a certain sensory effect after a certain action, a bidirectional association between neural representations of the action and the sensory effect is rapidly acquired, which enables action-driven prediction of the sensory effect. The present study aimed to test whether or not omission P3, an event-related brain potential (ERP) elicited by the sudden omission of a sensory effect, is sensitive to the formation of action-driven prediction. For this purpose, we examined how omission P3 is affected by the number of possible visual effects. In four separate blocks (1-, 2-, 4-, and 8-stimulus blocks), participants successively pressed a right button at an interval of about 1 s. In all blocks, each button press triggered a bar on a display (a bar with square edges, 85%; a bar with round edges, 5%), but occasionally did not (sudden omission of a visual effect, 10%). Participants were required to press a left button when a bar with round edges appeared. In the 1-stimulus block, the orientation of the bar was fixed throughout the block; in the 2-, 4-, and 8-stimulus blocks, the orientation was randomly varied among two, four, and eight possibilities, respectively. Omission P3 in the 1-stimulus block was greater than those in the 2-, 4-, and 8-stimulus blocks; there were no significant differences among the 2-, 4-, and 8-stimulus blocks. This binary pattern nicely fits the limitation in the acquisition of action-effect association; although an association between an action and one visual effect is easily acquired, associations between an action and two or more visual effects cannot be acquired concurrently. Taken together, the present results suggest that omission P3 is highly sensitive to the formation of action-driven prediction.