Several divergent cortical mechanisms generating multistability in visual perception have been suggested. Here, we investigated the neurophysiologic time pattern of multistable perceptual changes by means of a simultaneous recording with electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). Volunteers responded to the subjective perception of a sudden change between stable patterns of illusionary motion (multistable transition) during a stroboscopic paradigm. We found a global deceleration of the EEG frequency prior to a transition and an occipital-accentuated acceleration after a transition, as obtained by low-resolution electromagnetic tomography analysis (LORETA) analysis. A decrease in BOLD response was found in the prefrontal cortex before, and an increase after the transitions was observed in the right anterior insula, the MT/V5 regions and the SMA. The thalamus and left superior temporal gyrus showed a pattern of decrease before and increase after transitions. No such temporal course was found in the control condition. The multimodal approach of data acquisition allows us to argue that the top-down control of illusionary visual perception depends on selective attention, and that a diminution of vigilance reduces selective attention. These are necessary conditions to allow for the occurrence of a perception discontinuity in absence of a physical change of the stimulus.