Talk 1, 10:30 am
Geometry of anisotropic contextual interactions in the visual cortex places fundamental limits on spatial vision.
Mitchell Morton1,2 (), Sachira Denagamage1,2, Nyomi Hudson1, Anirvan Nandy1,2,3,4; 1Department of Neuroscience, Yale University, New Haven, CT 06510, 2Interdepartmental Neuroscience Program, Yale University, New Haven, CT 06510, 3Wu Tsai Institute, Yale University, New Haven, CT 06511, 4Kavli Institute for Neuroscience, Yale University, New Haven, CT 06511
Crowding, the impaired ability to accurately recognize a target stimulus among distractors, is a major bottleneck in visual perception. The spatial configuration of distractors in relation to the target profoundly influences perceptual fidelity. Notably, when a distractor is placed at a more eccentric point on the radial axis (termed ‘radial-out crowding’), it exerts the strongest impairment. Despite the pronounced perceptual anisotropy, the prevalent assumption underlying our understanding of contextual interactions in the visual cortex assumes isotropy. We investigated how distractor stimuli in different spatial configurations impacted the representation of a target stimulus in laminar microcircuits in the primary visual cortex (V1). Our study reveals that radial-out crowding more strongly impacts the ability to decode the target orientation from V1 population activity compared to other spatial configurations. This effect was strongest among putative excitatory neurons in the superficial and input layers, which are the primary neural populations involved in feed-forward information propagation. Remarkably, the feedback pathway involving the deep cortical layers does not exhibit anisotropy. Mechanistically, the anisotropy is explained by a tuned suppression and untuned facilitation of orientation responses, leading to an anisotropic broadening of tuning curves in the feedforward pathway, but not in the feedback pathway. These results underscore the non-uniform spatial integration of information by neurons in the visual cortex, establishing the presence of anisotropic contextual interactions in the earliest stages of cortical processing. By elucidating the distinct roles of feed-forward and feedback pathways in the context of crowding, this study advances our understanding of the intricate interplay between spatial arrangement, neural circuitry, and the constraints on perceptual fidelity during early visual processing.
Acknowledgements: NIH/NEI R01 EY032555, NARSAD Young Investigator Grant, Ziegler Foundation Grant, Yale Orthwein Scholar Funds, NIH/NINDS training grants T32-NS007224, T32-NS041228, and NIH/NEI core grant for vision research P30 EY026878
Talk 2, 10:45 am
Spatial configuration of contextual stimuli influences inter-laminar interactions in macaque primary visual cortex
Xize Xu1,2,4 (), Mitchell P. Morton1,3, Nyomi V. Hudson1, Anirvan S. Nandy1,3,4,5, Monika P. Jadi1,2,3,5; 1Department of Neuroscience, Yale University, New Haven, CT 06510, 2Department of Psychiatry, Yale University, New Haven, CT 06510, 3Interdepartmental Neuroscience Program, Yale University, New Haven, CT 06510, 4Kavli Institute for Neuroscience, Yale University, New Haven, CT 06511, 5Wu Tsai Institute, Yale University, New Haven, CT 06511
Our visual experience is a result of the concerted activity of neuronal ensembles in the sensory hierarchy. Yet how the spatial organization of objects influences neural activity in this hierarchy remains poorly understood. We investigate how the inter-laminar interactions in the primary visual cortex (V1) are affected by visual stimuli in isolation or with flanking stimuli at various spatial configurations that are known to exert a “crowding” effect on perception. Visual crowding is thought to be the primary limitation on object perception in peripheral vision, and the psychophysically identified “crowding zone” of impaired object identification is highly non-uniform. By employing dimensionality reduction approaches to simultaneous layer-specific population recordings, we determined the extent to which trial-to-trial fluctuations of population responses in the superficial layers of V1 are related to those in the input layer. We demonstrate that specific spatial configurations of contextual stimuli differentially modulate inter-laminar interactions by changing their fidelity and the balance between feedforward and feedback signaling, but not their structure. Remarkably, the modulations mirror the spatially non-uniform aspects of perceptual crowding. Our results suggest a model in which perceptual impairment under crowding is mediated by visual context integration in the superficial layers of V1 and posit that the non-uniformity in contextual inputs is the neural substrate of perceptual experience.
Acknowledgements: R01 EY032555; Swartz fellowship; Kavli fellowship
Talk 3, 11:00 am
Changes in population receptive fields following artificial scotoma; remapping or nonlinear responses?
Marcus Daghlian1,2,3,4 (), Mayra Bittencourt1,2,3,4, Remco Renken2,6, Serge Dumoulin1,3,4,5, Frans Cornelissen2,6; 1Spinoza Centre for Neuroimaging, 2Laboratory for Experimental Ophthalmology, UMCG, 3Netherlands Institute for Neuroscience, Royal Netherlands Academy of Sciences, 4Vrije Universiteit, 5Utrecht Universiteit, 6Rijksuniversiteit Groningen
There is considerable debate regarding the extent of plasticity in the adult visual cortex, following retinal lesions. Early electrophysiological studies in animal models provided evidence that receptive fields (RF) inside the lesion projection zone adaptively change position preference towards spared portions of the visual field. However, subsequent multimodal studies did not find evidence of RF position change. Changes in population RFs (pRF) have also been observed in healthy controls following simulated scotoma. Importantly, changes in pRF locations are not limited to the simulated lesion projection zone, as pRFs around the visual field display change their apparent position. This suggests that changes in pRFs around scotoma, either simulated or real, are not necessarily due to plasticity. Previous studies generally used a linear, single gaussian pRF model. However, the visual cortex responds non-linearly to stimuli, hence it is possible that apparent position changes following scotoma are driven by non-linear responses. To test this hypothesis, five participants were shown a standard retinotopic mapping stimulus, with and without simulated scotoma (an unstimulated mean-luminance patch on the screen). We modelled the pRF properties using both a linear single gaussian and a pRF model based on divisive normalization (DN). The DN pRF model captures non-linear responses, such as surround suppression, compression and oversaturation. In line with previous studies, we find changes in preferred position using the linear single gaussian pRF model. When fitting with the DN pRF model, the results also displayed position changes, however, these were significantly smaller. Our results suggest that at least a portion of pRF preferred position changes can be captured by non-linear responses. Thus, non-linear responses may be misinterpreted as signs of plasticity, and we propose that the studies of cortical plasticity and stability must consider non-linear responses of visual cortex which are part of normal cortical dynamics.
Talk 4, 11:15 am
Oculomotor influences on extrafoveal sensitivity
The human eyes are always in motion, alternating rapid gaze shifts (microsaccades) with slow smooth movements (drifts) even when attending to a single point. In the fovea, these fixational eye movements (FEM) have been shown to enhance sensitivity in complementary spatial frequency ranges, in a way that is consistent with their reformatting of spatial patterns into temporal signals: the luminance modulations from microsaccades and drifts emphasize low and high spatial frequencies, respectively. Outside the fovea, however, the perceptual roles of FEM remain unclear. In the periphery, views range from general functions (‘refreshing’ percepts) to no function because of the little FEM motion relative to receptive fields size. Here we show that FEM lead to similar perceptual consequences inside and outside the fovea. Human observers (N=6) were asked to report the orientation (±45°) of a full-field grating while maintaining fixation. The grating was either at high (10 cpd) or low (0.2 cpd) spatial frequency. To restrict stimulation to the peripheral visual field, a circular gray patch (diameter 15° or 23°) remained stationary on the retina centered on the line of sight. We compared performance in the presence and absence of the retinal motion caused by FEM. In the latter condition, eye movements were counteracted in real-time by moving the stimulus on the display via a custom apparatus. Our results show that, also outside the fovea, drifts and microsaccades selectively improve sensitivity to high and low spatial frequencies, respectively. On average performance dropped by approximately 10% at high frequency when retinal motion was eliminated and improved by a similar amount at low frequency in the trials with microsaccades. Together, these results indicate that FEM operate uniformly throughout the visual field, reformatting luminance patterns into spatiotemporal signals that enhance contrast sensitivity in complementary ranges of spatial frequencies.
Acknowledgements: This work was supported by NIH EY018363 and P30 EY001319
Talk 5, 11:30 am
Seeing less but seeing better: Information loss and accuracy gain in redundancy masking
To cope with excessive visual information in the environment, the visual system selects, discards, and compresses information. One compression mechanism is redundancy masking (RM) where redundant visual information is compressed. RM occurs with as few as three items. For example, when presented with three identical items in the visual periphery, observers often report seeing only two items. Here, we investigated to what extent features of masked items withstand or are lost in RM. We presented 3-5 radially arranged bars with varying widths (0.1°, 0.25°, 0.4°) for 150ms in the left or right hemifield (10° eccentricity). Observers reported the number of bars, and then adjusted probe widths and spacings to match the perceived stimulus. We computed deviation scores as the difference between perceived and actual (1) number of bars, (2) bar widths, (3) spacings between bars, and (4) overall widths of the arrays. There was strong RM: The number of bars reported was lower than the number presented. Overall, the width of thin (thick) bars tended to be overestimated (underestimated). In RM trials, the reported width was slightly larger than in non-RM trials. Importantly, except for the thinnest width condition, the reported width was more accurate in RM than in correct trials. The reported spacing between bars was larger in RM compared to correct trials, showing a lower perceived density in RM, while the reported overall extent of the arrays was smaller in RM trials, replicating previous results of visual space compression in RM. Our results suggest that the erroneous perception of smaller numbers of items in RM may go hand in hand with higher accuracy in reporting their features. We discuss how RM can be beneficial beyond the economical use of limited processing capacities by improving perception of individual items.
Acknowledgements: ANR-19-FRAL-0004; Tubitak 122N748
Talk 6, 11:45 am
Computational aspects of grouping explain visual crowding across space and time
Martina Morea1 (), Michael Herzog1, Gregory Francis2, Mauro Manassi3; 1Laboratory of Psychophysics, Brain Mind Institute, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland, 2Department of Psychological Sciences, Purdue University, West Lafayette, USA, 3School of Psychology, University of Aberdeen, King’s College, Aberdeen, United Kingdom
In crowding, perception of a target deteriorates in the presence of flanking elements. Surprisingly, crowding can sometimes be ameliorated when more flanking elements are presented, a phenomenon called uncrowding. It was previously shown that uncrowding occurs when the target ungroups from the flankers. Here, we show that grouping processes can explain not only spatial interactions in crowding, but also interactions in the time domain. First, we show that grouping requires a minimum stimulus duration to occur: when participants discriminated the offset of a target vernier presented alone or flanked by lines or cuboids, a cuboid duration of at least 160 ms was needed for uncrowding to occur. Second, we show that the grouping process can be initiated by presenting only the cuboids for 20 ms before an ISI and then the display with the cuboids and the target vernier for 20 ms. With the preview, uncrowding occurs for short ISIs of 20 ms up to ISIs of 250 ms, pointing out to recurrent grouping processes taking place. Third, when presenting flanking elements during the ISI, the uncrowding effect occurred only when the elements formed a good Gestalt. We show that this body of results can be well explained by the Laminart model, in which recurrent processing segments the visual scene into different objects, which are represented in separate segmentation layers. When the target and flanking cuboids are processed in different layers, uncrowding occurs; when they are in the same layer, crowding occurs. Importantly, the preview of the cuboids gives the model sufficient time to segment the vernier target away from the cuboids. Taken together, our results highlight the importance of recurrent grouping processes in spatial and temporal interactions in vision.
Talk 7, 12:00 pm
A continuous tracking measure of orientation sensitivity and bias in the visual periphery
Visual performance varies significantly across the visual field, revealing variations in sensitivity at different locations within and across observers. These include polar angle asymmetries—variations in performance across angular locations. Conventional methods for measuring these variations are time consuming but can be made more efficient with recent continuous tracking methods, in which observers follow a continuously changing target. This method calculates the peak of the cross-correlation between the tracked and reference stimuli, effectively assessing sensitivity. However, it does not directly quantify perceptual bias, which reflects systematic errors in perception. To address this, we introduce a novel approach to simultaneously map bias and sensitivity in orientation perception across the visual field at 8º eccentricity across four locations (upper, lower, right, and left). Participants fixate a central grating and adjust its orientation to match the orientation of a randomly rotating peripheral grating. We measured perceptual sensitivity by calculating the peak of the cross-correlation between the central and peripheral gratings. In addition, we measured bias by calculating the difference between observed and actual orientation values at each orientation, which are then grouped and averaged. To validate this approach, participants completed a second condition, in which we used the tilt illusion to measure biases in perceived orientation with a 45º annular surround for the peripheral grating. We reveal significant variations in the strength of the tilt illusion among participants and locations. Additionally, participants demonstrated significant variation in location-specific sensitivity in tracking the grating, both with and without the annulus in the periphery. Sensitivity was well correlated between the two tasks (p < .001), but lower with a surrounding annulus. Our results highlight individual differences in sensitivity and bias across the visual field with our novel continuous tracking paradigm, and variation in the magnitude of the tilt illusion in different peripheral field locations.
Acknowledgements: This work was supported by an NSERC Discovery Grant to AK