Background: In Zhaoping’s new framework for understanding vision, the central visual field enjoys feedback from higher to lower visual areas to aid visual recognition or discrimination, particularly in noisy and ambiguous situations. This recognition process is paraphrased as feedforward-feedback-verify-reweight (FFVW) as follows. First, forced by an information bottleneck, V1 feeds forward fragmentary, or information-reduced, signals to higher visual areas to suggest possible perceptual hypotheses about the visual scene/object; second, higher brain areas synthesize the would-be visual inputs consistent with each hypothesis according to its internal model of the visual world; third, the would-be inputs are fed back to lower visual areas such as V1 to compare with the actual inputs; fourth, a better match between the would-be and actual visual inputs increases the probability (weights) that the correspondingly responsible perceptual hypothesis is perceived as the perceptual outcome. This work investigates the feedback in the FFVW process in central vision using ambiguous depth perception in random-dot stereograms depicting a scene with a central disk in front of, or behind, a surrounding ring (see Figure).

Method: In a random-dot stereogram (RDS), the spatial disparities between the interocularly corresponding black and white random dots determine the depths of object surfaces. If a black dot in one monocular image corresponds to a white dot in the other (see the dots for the central disk in the lower RDS in the figure), disparity-tuned neurons in primary visual cortex (V1) respond as if their preferred disparities become non-preferred and vice versa, reversing the disparity sign reported to higher visual areas. Reversed depth is perceptible in the peripheral but not the central visual field. This study uses a novel paradigm by depicting a depth surface using a mixture of conventional contrast-match dots (for normal depth signals) and contrast-reversed dots (for reversed depth signals) for interocular correspondence. This mixing is applied to the central disk, while the surrounding ring uses only zero-disparity contrast-matched dots. Observers had to report whether the disk is in front of, or behind, the ring. The task is made difficult by additionally include interocularly uncorrelated noise dots in the monocular image areas for the disk.

Results: Adding contrast-reversed dots to a noisy RDS can augment or degrade perception of the depth surface of the central disk. Augmentation occurs when the reversed depth signals are congruent with the normal depth signals to report the same disparity sign (this is when the disparity of the contrast-reversed dots is the negative of the disparity of the contrast-matched dots), and occurs regardless of the viewing duration. Degradation occurs when the reversed and normal depth signals are incongruent with each other and when the RDS is viewed briefly. These phenomena reflect the Feedforward-Feedback-Verify-and-reWeight (FFVW) process for visual inference in central vision, and are consistent with the central-peripheral dichotomy that central vision has a stronger top-down feedback from higher to lower brain areas to disambiguate noisy and ambiguous inputs from V1. When a RDS is viewed too briefly for feedback, augmentation and degradation work by adding the reversed depth signals from contrast-reversed dots to the feedforward normal depth signals. With a sufficiently long viewing duration, the feedback vetoes incongruent reversed depth signals and amends or completes the imperfect, but congruent, reversed depth signals by analysis-by-synthesis computation.

Outlook: We know how disparity-tuned V1 neurons respond to normal and reversed depth signals, but we do not know how exactly feedback works from higher to lower visual areas. This study sets up a novel behavioural paradigm to dissect the feedforward-feedback processes, and this can be followed up by physiological studies in monkeys and in humans (e.g., in fMRI) to probe further and deeper the underlying algorithms and mechanisms.