Abstract
We used visual evoked potentials recorded in response to a perceptually multi-stable motion display to study neural correlates of phenomenal coherence. The display was a variation of the occluded diamond illusion (Lorenceau and Shiffrar, 1992). We presented a grid of orthogonal moving bars behind an occluding mask with a series of square apertures. The perception of the stimulus alternated between a coherent grid or separately moving groups of horizontal and vertical bars. Observers indicated with a button which perceptual state was dominant. The bars oscillated at slightly different temporal frequencies which allowed us to separately analyze the response to each component of the display at frequencies equal to the harmonics of the horizontal and vertical component motions. We also analyzed the response at frequencies equal to low order sums and differences between the two component frequencies. The EEG data was then separately processed for the perceptually coherent and incoherent states. Any changes in the evoked response were thus caused by changes in perceptual organization since the stimulus on the retina was constant. Across observers (n=9) the amplitude of the evoked response during perceptually coherent motion was similar to that recorded in the perceptually incoherent state. Individual observers showed idiosyncratic amplitude increases or decreases between the two perceptual states. However, the response phase at most of the frequencies measured was more similar across observers in the perceptually coherent state than in the perceptually incoherent state. Increased phase consistency in the coherent state across observers implies that the neural response becomes more stereotypic. The inconsistency of response phase across observers in the perceptually incoherent state may be due to the presence of several alternative perceptual organizations, each with their own corresponding state vector.