The Flip Tilt Illusion: Visible in Peripheral Vision as Predicted by the Central-Peripheral Dichotomy

Peripheral vision is mainly for looking rather than seeing

Vision includes looking and seeing. Looking, mainly via gaze shifts, selects a fraction of visual input information for passage through the brain's information bottleneck. The selected input is placed within the attentional spotlight, typically in the central visual field. Seeing decodes, i.e., recognizes and discriminates, the selected inputs. Hence, peripheral vision should be mainly devoted to looking, in particular, deciding where to shift the gaze. Looking is often guided exogenously by a saliency map created by the primary visual cortex (V1), and can be effective with no seeing and limited awareness. In seeing, peripheral vision not only suffers from poor spatial resolution, but is also subject to crowding and is more vulnerable to illusions by misleading, ambiguous, and impoverished visual inputs. Central vision, mainly for seeing, enjoys the top-down feedback that aids seeing in light of the bottleneck which is hypothesized to starts from V1 to higher areas. This feedback queries for additional information from lower visual cortical areas such as V1 for ongoing recognition. Peripheral vision is deficient in this feedback according to the Central-peripheral Dichotomy (CPD) theory. The saccades engendered by peripheral vision allows looking to combine with seeing to give human observers the impression of seeing the whole scene clearly despite inattentional blindness.

Top-down generation of low-resolution representations improves visual perception and imagination

Perception or imagination requires top-down signals from high-level cortex to primary visual cortex (V1) to reconstruct or simulate the representations bottom-up stimulated by the seen images. Interestingly, top-down signals in V1 have lower spatial resolution than bottom-up representations. It is unclear why the brain uses low-resolution signals to reconstruct or simulate high-resolution representations. By modeling the top-down pathway of the visual system using the decoder of a variational auto-encoder (VAE), we reveal that low-resolution top-down signals can better reconstruct or simulate the information contained in the sparse activities of V1 simple cells, which facilitates perception and imagination. This advantage of low-resolution generation is related to facilitating high-level cortex to form geometry-respecting representations observed in experiments. Furthermore, we present two findings regarding this phenomenon in the context of AI-generated sketches, a style of drawings made of lines. First, we found that the quality of the generated sketches critically depends on the thickness of the lines in the sketches: thin-line sketches are harder to generate than thick-line sketches. Second, we propose a technique to generate high-quality thin-line sketches: instead of directly using original thin-line sketches, we use blurred sketches to train VAE or GAN (generative adversarial network), and then infer the thin-line sketches from the VAE- or GAN-generated blurred sketches. Collectively, our work suggests that low-resolution top-down generation is a strategy the brain uses to improve visual perception and imagination, which inspires new sketch-generation AI techniques.

Peripheral and central sensation: multisensory orienting and recognition across species

Attentional bottlenecks force animals to deeply process only a selected fraction of sensory inputs. This motivates a unifying central-peripheral dichotomy (CPD), which separates multisensory processing into functionally defined central and peripheral senses. Peripheral senses (e.g., human audition and peripheral vision) select a fraction of the sensory inputs by orienting animals' attention; central senses (e.g., human foveal vision) allow animals to recognize the selected inputs. Originally used to understand human vision, CPD can be applied to multisensory processes across species. I first describe key characteristics of central and peripheral senses, such as the degree of top-down feedback and density of sensory receptors, and then show CPD as a framework to link ecological, behavioral, neurophysiological, and anatomical data and produce falsifiable predictions.

The central-peripheral dichotomy and metacontrast masking

According to the central-peripheral dichotomy (CPD), feedback from higher to lower cortical areas along the visual pathway to aid recognition is weaker in the more peripheral visual field. Metacontrast masking is predominantly a reduced visibility of a brief target by a brief and spatially adjacent mask when the mask succeeds rather than precedes or coincides with the target. If this masking works mainly by interfering with the feedback mechanisms for target recognition, then, by the CPD, this masking should be weaker at more peripheral visual locations. We extended the metacontrast masking at fovea by Enns and Di Lollo to visual field eccentricities 1∘, 3∘, and 9∘. Relative to the target’s onset, the mask appeared at a stimulus onset asynchrony (SOA) of −50, 0, 50, 92, or 142 milliseconds (ms). Enlarged stimuli were used for larger eccentricities to equalize target discrimination performance across eccentricities as best as possible for zero SOA and when SOA was too long for substantial masking. At each eccentricity, the masking was weakest at 0 or −50 ms SOA, strongest at 50 ms SOA, and weakened with larger (positive) SOAs. Consistent with the CPD, larger eccentricities presented weaker maskings at all nonzero, and particularly the positive, SOAs.

Central-peripheral dichotomy: color-motion and luminance-motion binding show stronger top-down feedback in central vision

Recently a theory (Zhaoping, Vision Research, 136, 32-49, 2017) proposed that top-down feedback from higher to lower visual cortical areas, to aid visual recognition, is stronger in the central than in the peripheral visual fields. Since top-down feedback helps feature binding, a critical visual recognition process, this theory predicts that insufficient feedback in the periphery should make feature misbinding more likely. To test this prediction, this study assessed binding between color and motion features, or between luminance and motion features, at different visual field eccentricities. We first used color-motion stimuli containing equiluminant red and green dots moving in opposite directions, for example, red dots moved leftward while green dots moved rightward. Such stimuli were shown in both a central reference strip and a peripheral test strip; participants reported whether it was the first or second interval in a trial in which the dots of each color moved in the opposite directions between the two strips. The center of the test strip was at 4° or 15° away from the gaze fixation. Participants' performance was much worse when the test strip was more peripheral, suggesting that feature misbinding occurred more frequently there. This held even when the size and density of the dots were adjusted by eccentricity-dependent cortical magnification factors, and even when red/green dots were replaced by yellow/blue dots or black/white dots to suit the retinal input sampling peripherally. Our findings support that top-down feedback is more directed to central vision, which can resolve ambiguities in feature binding at more central visual locations.

Parallel Advantage: Further Evidence for Bottom-up Saliency Computation by Human Primary Visual Cortex

Finding a target among uniformly oriented non-targets is typically faster when this target is perpendicular, rather than parallel, to the non-targets. The V1 Saliency Hypothesis (V1SH), that neurons in the primary visual cortex (V1) signal saliency for exogenous attentional attraction, predicts exactly the opposite in a special case: each target or non-target comprises two equally sized disks displaced from each other by 1.2 disk diameters center-to-center along a line defining its orientation. A target has two white or two black disks. Each non-target has one white disk and one black disk, and thus, unlike the target, activates V1 neurons less when its orientation is parallel rather than perpendicular to the neurons' preferred orientations. When the target is parallel, rather than perpendicular, to the uniformly oriented non-targets, the target's evoked V1 response escapes V1's iso-orientation surround suppression, making the target more salient. I present behavioral observations confirming this prediction.