Abstract
Defining the neural correlates of awareness has been difficult due to the lack of selective means to measure consciousness-dependent activity. A widely used paradigm to explore visual awareness is binocular rivalry (BR): conflicting monocular images alternate every few seconds, with the dominant stimulus transiently gaining access to consciousness, while the non-dominant becomes invisible. From single-unit recordings in primates to whole brain EEG or fMRI recordings in humans, current techniques have only found moderate modulations of neural activity when stimuli are dominant versus when they are not; all techniques systematically reveal a significant response to invisible stimuli. SWIFT (semantic wavelet-induced frequency-tagging) is a recently developed EEG technique which selectively tracks high-level representations with high temporal resolution (Koenig-Robert & VanRullen, VSS 2011). SWIFT uses cyclic wavelet-scrambling to modulate the semantic content of an image at a temporal frequency f0, while conserving low-level features. We used SWIFT to measure perceptual alternations during BR. Two face pictures, one red and one green, were modulated under SWIFT at different f0 (1.4894 and 2 Hz) and presented monocularly. Two kinds of alternations were used: BR, where the two stimuli were presented simultaneously (and only one was perceived at a time), and physical alternation (PA), where the stimuli were presented alternately in each eye, mirroring the dynamics of BR spontaneous alternations. Subjects (n=9) reported the color of the perceived face. Fronto-central and occipito-temporal EEG activities were elicited by dominant stimuli, and were comparable in their topography and amplitude in PA and BR conditions. Crucially, non-dominant stimuli in BR elicited no more activity than the response to a blank screen in PA. SWIFT thus makes it possible to non-invasively record brain signals that are not merely modulated by BR, but follow its perceptual alternations in an all-or-none fashion, exactly as their phenomenal counterpart does.
Meeting abstract presented at VSS 2012