Top-down influences on bistable perception revealed by event-related potentials

Electrophysiological recordings were made in 27 observers to investigate whether differences in recently identified ERP components (reversal positivity, reversal negativity, late positive component) would be revealed during naturally occurring perceptual reversals of a bistable figure. Additionally, to investigate the role of top-down control over perceptual reversals, observers were instructed to maintain three distinct intentional approaches in separate conditions: (1) 'speedup' reversals, (2) 'hold' one of the two percepts stable, and (3) remain passive and allow reversals to occur naturally. Using a lattice of Necker cubes as a stimulus, Kornmeier and Bach (2004, 2005) reported early amplitude differences related to perceptual reversals, the reversal positivity at 120msec post-stimulus, the reversal negativity at 180-380msec, and the late positive component at 30-500msec. The current study investigated whether these event-related potentials (ERPs) of Necker cube reversals (using the same stimulus as Kornmeier and Bach) could be identified using more conservative ERP analysis techniques, and if so, where on the scalp these components could be recorded. Additionally, intention-based ERP differences were assessed by comparing P1, N1, P2, N2, and reversal negativity amplitudes across the three intention conditions. Results revealed the broad reversal-related negativity at right posterior electrode sites and the late positive component at central parietal sites, but the reversal positivity could not be identified. When perceptual reversals occurred, intention-based differences were found for the P2, N2, and reversal negativity, in which amplitudes were more negative for the intention-to-reverse condition compared to the passive condition. An N2 enhancement was also found for unintentional reversals compared to passive reversals. A trend toward amplitude enhancement of the N1 component was found for the intention-to-reverse condition compared to both of the other conditions, especially for observers who were successful at the intention task. These results support a theory of bistable perception in which changes in early spatial attention (indicated by N1 enhancement) lead to changes in selective attention (indicated by the reversal negativity), which determine how one perceives the bistable figure. Evidence from this experiment also suggests that these same attention mechanisms can be modulated by top-down intentional control. A neuroanatomical model is presented to conceptualize these findings.