Abstract
Some phenomena, e.g., visual crowding (the reduced capability to recognize an object in visual clutter), are stronger in the peripheral visual field. Given the attentional bottleneck, which massively reduces visual input information flow starting from the primary visual cortex (V1) to higher visual cortical areas (Zhaoping, Current Opinion in Neurobiology, 2019), I recently proposed (Zhaoping, Vision Research, 2017) that the top-down feedback from higher to lower visual cortical areas for recognition is weaker or absent in the peripheral visual field, since attended objects are typically brought to the central visual field. This feedback facilitates analysis-by-synthesis-based object recognition, it involves assessing the agreement between the brain's internal model of the visual inputs for a visual object and the actual visual input in lower visual areas such as V1, where the relevant sensory input has yet to be filtered out by the attentional bottleneck, so that a good or poor agreement makes the object more or less likely perceived. This feedback analysis is particularly helpful for recognition when visual inputs are noisy or uncertain.
This proposal successfully predicted (Zhaoping and Ackermann, Perception, 2018) that, unlike vision in the central visual field, vision in the peripheral visual field can perceive reversed depth in a binocularly anticorrelated random-dot stereogram (aRDS), in which a bright dot in one eye corresponds to a dark dot in the other eye and vice versa, and which excites V1 neurons tuned to the binocular spatial disparity opposite to the disparity between the corresponding dots (Cumming and Parker, Nature, 1997).
I propose that the prevalent visual field location, i.e., central or peripheral, for any visual phenomenon is diagnostic for the feedforward or feedback neural mechanisms responsible. Accordingly, the reduction of feedforward visual information from V1 causes crowding unless the feedback analysis compensates sufficiently; the reversed depth in an aRDS is invisible in the central visual field since the feedback vetoes the feedforward sensory inputs which violate the brain's internal model that the inputs should be binocularly correlated. Further, in visual backward masking, perception of a briefly presented target is disrupted by a mask presented shortly after, but not before, the target. If this is caused by a conflict between the top-down feedback for the target and the feedforward input from the mask, then I predict that the masking effect is weaker in peripheral visual field. We discuss the central-peripheral diagnostic in relation to other visual phenomena, including visual illusions.
