Biology subjects

Sayim, B.

Publications and source records attributed to Sayim, B..

3 recordsLinked to original sources

bioRxiv2026

Compression of redundant visual information improves feature discrimination in human vision

The visual environment contains more information than can be fully processed. To cope with this information, the visual system selects, integrates, and compresses sensory inputs. These processes are particularly evident in peripheral vision, where nearby items strongly limit access to individual elements, as in crowding and redundancy masking. In redundancy masking, repeated items are compressed into fewer items than are physically present, for example, three spatially separated items presented in the periphery are often perceived as only two items. Whether these interactions among items merely limit access to visual information or serve a functional role in perception remains unknown. Here, we investigated whether redundancy masking, despite reducing the number of perceived items, improves feature discrimination. In two experiments, participants viewed arrays of identical bars in peripheral vision, reported the number of bars they perceived, and adjusted a foveal probe to match perceived width (Experiment 1) or width and spacing (Experiment 2). Strong redundancy masking occurred in both experiments. When participants perceived fewer bars than were presented, perceived width was closer to the presented bar width than when all bars were perceived, where width was systematically underestimated. In Experiment 2, the perceived spacing between the (fewer) perceived bars increased. We modeled feature discrimination under redundancy masking and found that the observed improvement could not be explained by averaging features of the lost and perceived bars. These results suggest that the visual system efficiently compresses redundant information, reducing the number of perceived items while improving feature discrimination.

animal behavior and cognition↗

bioRxiv2025

Redundancy masking and the compression of information in the brain

The visual world is inherently complex, presenting far more information than a human visual system can process in full. To manage this overload, the visual brain employs several mechanisms. One mechanism that possibly contributes to the reduction of information is redundancy masking (RM): the reduction of the number of perceived items in repeating patterns. For example, when three identical lines are presented in the periphery, observers often perceive only two. The underlying neural mechanisms of RM remain unclear. Here, we use steady-state visual evoked potential (SSVEP) to examine whether redundancy-masked items are neurally suppressed or integrated with neighboring items. Three identical arcs (quarter-circles; 0.44{degrees} line width) were presented in the periphery (eccentricities: 17.3{degrees}, 19.5{degrees}, and 21.7{degrees}), each tagged with a unique frequency. Participants maintained central fixation, monitored via a gaze-contingent control, and reported the number of arcs they predominantly perceived after each 10s trial. We analyzed baseline-corrected amplitudes at each tagged frequency and calculated signal-to-noise ratios (SNRs) for fundamental and intermodulation (IM) components, separating trials by behavioral responses (RM: 2 items perceived, non-RM: 3 items perceived). Fundamental frequency comparisons revealed that the outer arc elicited higher SSVEP responses than the inner and middle one under RM, with no significant differences between arcs under non-RM. However, fundamental frequency SNRs did not differ between RM and non-RM perceptions. When we compared IM SNRs, the middle and outer arcs combination was significantly higher during RM compared to non-RM, suggesting increased neural integration between them. These results indicate that RM involves a loss of conscious access to visual information, yet corresponding neural signals are not entirely suppressed. Instead, the neural signatures we found suggest the integration with neighboring elements across space and time. We suggest that redundancy-masked items - although unavailable for conscious report- are still observed in the neural signatures of RM.

neuroscience↗

bioRxiv2019

Capsule Networks but not Classic CNNs Explain Global Visual Processing

Classically, visual processing is described as a cascade of local feedforward computations. Feedforward Convolutional Neural Networks (ffCNNs) have shown how powerful such models can be. However, using visual crowding as a well-controlled challenge, we previously showed that no classic model of vision, including ffCNNs, can explain human global shape processing (1). Here, we show that Capsule Neural Networks (CapsNets; 2), combining ffCNNs with recurrent grouping and segmentation, solve this challenge. We also show that ffCNNs and standard recurrent CNNs do not, suggesting that the grouping and segmentation capabilities of CapsNets are crucial. Furthermore, we provide psychophysical evidence that grouping and segmentation are implemented recurrently in humans, and show that CapsNets reproduce these results well. We discuss why recurrence seems needed to implement grouping and segmentation efficiently. Together, we provide mutually reinforcing psychophysical and computational evidence that a recurrent grouping and segmentation process is essential to understand the visual system and create better models that harness global shape computations. Author SummaryFeedforward Convolutional Neural Networks (ffCNNs) have revolutionized computer vision and are deeply transforming neuroscience. However, ffCNNs only roughly mimic human vision. There is a rapidly expanding body of literature investigating differences between humans and ffCNNs. Several findings suggest that, unlike humans, ffCNNs rely mostly on local visual features. Furthermore, ffCNNs lack recurrent connections, which abound in the brain. Here, we use visual crowding, a well-known psychophysical phenomenon, to investigate recurrent computations in global shape processing. Previously, we showed that no model based on the classic feedforward framework of vision can explain global effects in crowding. Here, we show that Capsule Neural Networks (CapsNets), combining ffCNNs with recurrent grouping and segmentation, solve this challenge. ffCNNs and recurrent CNNs with lateral and top-down recurrent connections do not, suggesting that grouping and segmentation are crucial for human-like global computations. Based on these results, we hypothesize that one computational function of recurrence is to efficiently implement grouping and segmentation. We provide psychophysical evidence that, indeed, grouping and segmentation is based on time consuming recurrent processes in the human brain. CapsNets reproduce these results too. Together, we provide mutually reinforcing computational and psychophysical evidence that a recurrent grouping and segmentation process is essential to understand the visual system and create better models that harness global shape computations.

neuroscience↗