Spatial Vision: Crowding and eccentricity, clinical, models

Talk Session: Sunday, May 18, 2025, 8:15 – 10:00 am, Talk Room 1

Schedule of Events | Search Abstracts | Symposia | Talk Sessions | Poster Sessions

Talk 1, 8:15 am

Mechanisms of foveal crowding: insights from retinal imaging and retinal-contingent psychophysics

Krishnamachari Prahalad¹ (), Ashley M. Clark¹, Benjamin Moon¹, Austin Roorda², Pavan Tiruveedhula², Wolf Harmening³, Aleksandr Gutnikov³, Samantha K. Jenks¹, Sanjana Kapisthalam¹, Michele Rucci¹, Jannick P. Rolland¹, Martina Poletti¹; ¹University of Rochester, ²University of California, Berkley, ³University of Bonn

Visual crowding, the interference in recognition of a stimulus caused by surrounding similar objects, occurs not only in the periphery but also at the center of gaze, where visual resolution is highest. Yet, the mechanisms underlying foveal crowding are still unclear, due to the difficulty in controlling for optical factors and fixational eye motion. To address these challenges, we used an Adaptive Optics Scanning Light Ophthalmoscope (AOSLO) to stimulate the retina while reducing the impact of optical aberrations and maintaining the stimulus at a fixed retinal location despite the presence of fixational eye movements. Subjects (N=8) performed a 4AFC digit identification task using Pelli’s font, designed specifically to probe foveal crowding. Stimulus sizes were set at three times the acuity threshold. Flanker distances varied using the method of constant stimuli. Stimuli were presented under retinal stabilization at the Preferred Retinal Locus (PRL) and 15 arcmin away. By assessing cone spacing at the stimulated location, our results show that the spatial extent of crowding at the PRL approximates a single cone diameter (1.15 ± 0.48 times the cone diameter). However, just 15 arcmin away the extent of crowding exceeds cone spacing (3.08 ± 2.08 times the cone diameter), revealing that additional cortical pooling mechanisms dominate with increasing eccentricity even within the foveola. Further, at the PRL, incorrect responses showed no bias in reporting either flanker, but 15 arcmin away, subjects were more likely to report the inner flanker on incorrect trials. These findings reveal that photoreceptors spacing predicts the spatial extent of crowding only at the PRL. They indicate that further signal pooling takes place just a few arcmin away from this location suggesting the presence of a cortical magnification gradient within the fovea.

R01 EY029788, R01 EY018363, EY001319, R01 EY023591, Ha5323/6-1 and Ha5323/8-1

Talk 2, 8:30 am

Perceptograms measured for Amblyopic form distortions, modeled with cortical deficits

Akihito Maruya¹ (), Farzaneh Olianezhad¹, Jingyun Wang¹, Jose-Manuel Alonso¹, Qasim Zaidi¹; ¹State University of New York, College of Optometry

Striking form distortions seen by amblyopes have been documented by showing high contrast sinusoidal gratings to the amblyopic eye (AE) and drawing the percepts through the fellow eye (FE). The distortions fall into 7 classes, each resembling sums-of-grating-pairs (Barrett et al 2003). These distortions of orientation processing are keys to understanding neural deficits in amblyopic cortex. Drawings may not accurately represent distortions of contrast, shading and frequency, so we measured perceptograms for 4 amblyopes ages 22-45. In a dichoptic display, AE was shown a 3° test grating in the center (6, 9, 12 cyc/deg, 4 orientations, ON or OFF). FE was shown 8 surrounding patterns: the test grating and the 7 types of distortion plaids. The observer picked the pattern most like the central image. Then the AE was shown the test grating and the FE the chosen match while contrast, frequency, orientation and phase of the 2 constituent gratings were adjusted until the two percepts matched perfectly (3 repeats). In the model, images were convolved with the optical point-spread-function, passed through the ON/OFF nonlinearity, and convolved with center-surround retino-thalamic filters with multiplicative internal noise. Then a formal equation that the signals generated in visual cortex by the test grating seen through AE match the signals generated by the matched perceptogram seen through FE, was used to analytically derive the modified cortical filters processing AE signals for 24 perceptograms jointly for each observer, using standard orientation tuned filters for normal cortex. The modified filters recreated the distortions seen through AE and were appreciably broader in tuning, slightly shifted in orientation preference, and varied in response magnitude. The orientation-based changes in neuronal tuning-width and response-magnitude provide a target for neural development models of amblyopia and could predict distortions in other orientation-based percepts such as contours, 3D shape-from-texture, mirror-symmetry and object poses.

NEI grants EY035085 & EY035838

Talk 3, 8:45 am

Localization biases in the periphery are idiosyncratic: Evidence from over 9000 observers

Anna Kosovicheva¹ (), Ido Ziv Li¹, Jihahm Yoo², Jeremy M. Wolfe^3,4, Jiali Song¹; ¹University of Toronto Mississauga, ²Korea Science Academy of KAIST, ³Brigham & Women’s Hospital, ⁴Harvard Medical School

Accurately registering the location of an object is a fundamental visual process. Previous studies have emphasized commonalities across individuals in the effect of the polar angle of the target relative to fixation in visual localization. However, there is also evidence that individuals exhibit consistent, idiosyncratic patterns of directional, angular error when reporting target locations in the periphery, and such patterns of error are weakly correlated between observers. This evidence comes from small-scale laboratory studies, involving dozens of participants, which may be underpowered to detect subtle consistencies across the population. We examined the consistency of individuals’ localization errors using a large-scale dataset from an online game (over 9,400 observers and 4.5 million trials across 639,000 sessions). On each trial, participants were instructed to identify a symbol in the center of the screen and were simultaneously shown a peripheral target. The target could appear with 0-10 distractor items. The eccentricities of the peripheral targets and distractors varied randomly and independently across trials. Participants clicked on the location of the peripheral target, and if correct, were asked to identify the central symbol among 5 alternatives. We analyzed trials where participants correctly identified the symbol and localized the target. We divided trials into bins based on their polar angle. We then calculated pairwise correlations in the angular (directional) click error between participants relative to display center (clockwise vs. counterclockwise). We found that directional localization errors were, on average, uncorrelated between all possible pairs of participants. Between-subject correlations of localization errors were normally distributed and centered around 0. However, within individuals, errors were non-random. Split-half correlation yielded a reliable, positive correlation (r = 0.41, t(9411)= 185.82, p< 0.001). These results align with the findings of small-scale laboratory studies and suggest that consistent idiosyncratic localization errors in the visual periphery are uncorrelated at the population level.

This work was supported by an NSERC Discovery Grant to AK, and NIH EY017001 and NSF 2146617 to JMW.

Talk 4, 9:00 am

Where internal noise and efficiency underlie visual field asymmetries

Shutian Xue¹ (), Antoine Barbot¹, Rachel Chen¹, Marisa Carrasco¹; ¹New York University

[Background] Visual performance typically peaks at the fovea and declines with eccentricity, and exhibits polar angle asymmetries: better performance along the horizontal than the vertical meridian and better performance at the lower than the upper vertical meridian. Do these performance differences reflect differential ability to process task-relevant information from noise? Here, we investigate how two factors limiting performance–internal noise (amount of internal variability in the system) and efficiency (the ability to extract information from the target)–underlie performance differences throughout the visual field. [Method] At each location, observers discriminated the orientation (±45º off the vertical axis) of Gabor patches (4, 5 and 6 cpd) embedded in varying contrast levels of dynamic white noise. Using an equivalent noise protocol, we mapped contrast threshold as a function of noise contrast to estimate additive and multiplicative internal noise, as well as efficiency, at three eccentricities (fovea, parafovea: 4°, perifovea: 8°) and four polar angles (left and right horizontal meridian, upper and lower vertical meridian). [Results] (1) Additive (but not multiplicative) internal noise increased with eccentricity. Efficiency was higher at the fovea and 4º than 8º eccentricity, higher at the horizontal than vertical meridian at 8º eccentricity, and at the lower- than upper- vertical meridian at both 4º and 8º eccentricities. (2) Multiplicative (but not additive) internal noise was similar at 4 and 5 cpd and lowest at 6 cpd. Efficiency was higher at 4 cpd than at 5 and 6 cpd. These differences were similar across locations. [Conclusion] Distinct computations limit performance throughout the visual field: Additive internal noise primarily underlies eccentricity differences, consistent with variations in cortical surface area. Efficiency primarily underlies polar angle asymmetries, particularly in the perifovea, reflecting variation in neural tuning properties.

Funding: NIH R01-EY027401 to M.C.; NIH training grant to NYU, NIH1F31EY036732-01 to S.X.

Talk 5, 9:15 am

The neural signatures of redundancy masking investigated by EEG frequency tagging

Nihan Alp¹, Dogukan Nami Oztas¹, Li L-Miao², Bilge Sayim²; ¹Sabanci University

Redundancy masking (RM) –the reduction of the number of perceived items in repeating patterns– occurs with as few as three items. For example, when three identical, closely spaced lines are presented in the periphery, individuals often perceive only two lines. The underlying neural mechanisms of these substantial detection-like errors remain elusive. Here, we use steady-state visual evoked potential to examine the neural correlates of RM. Three arcs (quarter-circles; 0.44° line width) were presented for 10s either on the right or left side of fixation at 17.3°, 19.5° and 21.7° eccentricity. Each arc was tagged with a different frequency. Participants were instructed to maintain fixation, and the stimulus disappeared if participants' gaze shifted using a gaze-contingent paradigm. After the stimulus offset, participants reported the number of arcs they mostly perceived during a trial. We quantified baseline-corrected amplitudes for each arc (tagged frequency and harmonics) and calculated signal-to-noise ratios (SNRs) for intermodulations, separating both by the behavioral responses (RM: 1 or 2, non-RM: 3). The middle arc elicited a weaker amplitude compared to the inner one, with no significant differences between the middle and outer or the inner and outer arcs. Across intermodulations, SNRs for response ‘2’ were significantly higher than ‘3’, indicating greater neural integration when perceiving 2 arcs (RM) than 3 (non-RM). Specifically, the integration of inner and middle arcs, as well as middle and outer arcs, indicated by corresponding intermodulations, was significantly stronger when participants perceived 2 arcs compared to 3, further supporting stronger integration when RM occurred. These results indicate that while RM involves a loss of access to visual information, the lost signals may nevertheless be integrated by the visual system across space and time. We suggest that the effects of redundancy-masked items –although unavailable for conscious report– are still observed in the neural signatures of RM.

ANR-19-FRAL-0004; TUBITAK 122N748

Talk 6, 9:30 am

Probability summation model of spatial pooling within the cardinal axes of DKL color space

Christopher S Wu¹, Daniel R Coates¹; ¹University of Houston

The critical area of spatial summation is thought to reflect the spatial extent of some pooling mechanism in the visual system, although, the exact underlying processes are unknown. Previously, we have characterized the spatial summation of chromatic contrasts defined by the cardinal axes of DKL color space. Here, we describe a simple model to bridge physiological properties of the visual system and functional findings. Spatial summation functions were obtained for contrasts along the cardinal directions of DKL color space in four subjects. Functions were measured in the oblique meridians at 5, 10, 15, and 20 degrees eccentricity with stimuli consisting of solid circular spots of multiple sizes. Contrast detection thresholds to images of these stimuli were simulated through the model. The model simulates the pooling of information through the visual system using a cascade of convolutional filtering and downsampling layers. Filtering kernels representing receptive fields were generated based on eccentricity dependent sizes and the spatial overlap of specific retinal ganglion cell type dendritic fields. Relative detection thresholds were calculated through probability summation. The model was able to reproduce the bilinear shape characteristic of human spatial summation functions, matching our psychophysical dataset across eccentricities and within theorized visual processing pathways (average r2 across all conditions = 0.71). The resultant critical areas matched the size of simulated receptive fields, while the slope of partial summation could be controlled by the exponent of probability summation. The average exponent of summation was 4.0 in the achromatic condition, 1.6 in the +(L-M), 1.4 in the -(L-M), 3.1 in the +S-(L+M), and 2.7 in the -S-(L+M) conditions. This fitted exponent likely reflects the processing characteristics of pathway-dependent higher level cortical mechanisms. Thus, a combination of probability summation and physiologically inspired filtering can account for the distinct psychophysical spatial summation functions obtained across eccentricities and pathways.