Binocular Rivalry and Perceptual Filling-in of Visual Phantoms in Human Visual Cortex

M. Meng,F. Tong
DOI: https://doi.org/10.1167/4.8.63
2004-01-01
Journal of Vision
Abstract:During binocular rivalry, observers can perceive only one of two competing monocular stimuli at any time. In contrast, visual phantoms involve perceptual filling-in across a gap between two collinearly aligned gratings. Here, we used psychophysics and fMRI to investigate the interactions of the neural mechanisms mediating these two phenomena. Such interactions can potentially reveal the sequence of processing in the visual pathway. We devised a novel display, which revealed that rivalry determines the outcome of phantom filling-in. One eye was shown two vertically aligned gratings, positioned one above the other, which would normally lead to perception of a visual phantom in the gap, while the other eye viewed two horizontal gratings that would normally not lead to a visual phantom. During binocular rivalry between these stimuli, subjects reported phantom filling-in only when the vertically aligned gratings were dominant. These results suggest that binocular rivalry occurs at an earlier stage of processing than phantom filling-in. We also monitored fMRI activity while subjects viewed this perceptually ambiguous display. When subjects perceived the vertically aligned gratings as dominant, increased activity was observed in retinotopic regions corresponding to the phantom location in areas V1 and V2. In contrast, when the horizontal gratings became dominant and observers failed to perceive a phantom, decreased activity was found. Our results indicate that: 1) rivalry can suppress the formation of visual phantoms and therefore likely occurs at an earlier stage of visual processing, 2) neural activity in V1 and V2 reflects the observer's conscious perception of the visual phantom suggesting filling-in occurs in these early visual areas, and 3) assuming that rivalry precedes filling-in, it follows that binocular rivalry likely occurs at a processing stage no later than V1 (cf. Polonsky et al., 2000; Tong & Engel, 2001; but see also Leopold & Logothetis, 1996). NIH grants R01-EY14202, P50-MH62196
What problem does this paper attempt to address?