BAVRD 2024 Schedule

Time	Agenda
9:00-9:30	Check in (Breakfast Served)
9:30-9:35	Introduction
9:35-10:00	Seoyoung Ahn, PhD, HHMI/University of California, Berkeley Title: Beyond a single glance: Modeling active sampling of objects and scene using DNNs
10:00-10:10	LT1: Mark Lescroart, PhD, University of Nevada, Reno Title: Scene- and object-based tasks performed on the same complex stimuli activate different regions in parietal and lateral occipital cortex
10:10-10:20	LT2: Patricia Limon, BS, Stanford University Title: Effect of Attentional State on Experience Dependent Visual Cortical Neuroplasticity: Contrast Response Function Changes as a Result of Attention to Visual Stimulus Across Days
10:20-10:45	Katrin Franke, PhD, Stanford University School of Medicine Title: Mapping neuronal representations in the visual cortex using deep learning
10:45-10:55	LT3: YuYi Lu, BS, BA, University of California, Berkeley Title: The locus of redundancy masking
10:55-11:05	LT4: Ryan Ash, MD, PhD, Stanford University School of Medicine Title: Noninvasive neuromodulation of human visual thalamus with transcranial ultrasound stimulation
11:05-11:15	Break (Coffee & Tea Served)
11:15-11:40	Michael Silver, PhD, University of California, Berkeley Title: Predictions and visual perceptual selection
11:40-11:50	LT5: Wendy Torrens, MS, University of Nevada, Reno Title: Anomalous visual cortical excitability in Schizotypy
11:50-12:00	LT6: Vaidehi Natu, PhD, Stanford University Title: Infants' cortex undergoes microstructural growth coupled with myelination during development
12:00-12:50	Keynote: Dan Yamins PhD, Stanford University Title: Climbing the Ladder of Causation Toward Intermediate Vision
12:50-1:50	Lunch Break (Lunch Served)
1:50-2:15	Jorge Otero-Millan, PhD, University of California, Berkeley Title: Learning to hold the eyes still and aligned
2:15-2:25	LT7: Reem Almagati OD, MSc, University of California, Berkeley Title: Effects of Convergence and Monocular viewing on Perception of the World's Horizontal and Vertical in Virtual Reality
2:25-2:35	LT8: Orhan Soyuhos, MSc, University of California Davis Title: Posterior Parietal Inactivation Alters Intrinsic Neural Timescales and Attentional Processing in Frontal Eye Field Neurons
2:35-2:45	Break (Coffee & Tea Served)
2:45-3:10	Tyson Kim, MD, PhD, UCSF School of Medicine Title: Transscleral Multiphoton Imaging and Flowmetry
3:10-3:20	LT9: Kiara Abhayaratne, University of California, Berkeley Title: Modeling Photoreceptor Loss and Behavioral Performance in Retinitis Pigmentosa in rd10 Mice Using a Cricket-Hunting Assay
3:20-3:30	LT10: Yujia Yang, MD, PhD, University of California, Berkeley Title: Reduced corneal susceptibility to colonization by commensal bacteria after contact lens wear is associated with parainflammation involving Lyz2+ cell trafficking toward the corneal surface
3:30-3:35	Ending message
3:35-5:00	Poster Session + Drinks Reception (Beverages & Bites Served)

Poster Sessions

Click here to see details about the poster presentations.

Speaker Abstracts

Seoyoung Ahn, PhD

Institution: HHMI/University of California, Berkeley
Talk Title: Beyond a single glance: Modeling active sampling of objects and scene using DNN
Learn More: Google Scholar

Mark Lescroart, PhD

Institution: University of Nevada, Reno
Talk Title: Scene- and object-based tasks performed on the same complex stimuli activate different regions in parietal and lateral occipital cortex
Learn More: Google Scholar

Abstract

Visual tasks activate regions throughout the parietal, frontal, and inferior temporal cortex. Past work has found that many of these regions are similarly activated by a wide variety of tasks and consequently these regions together have been called the Multiple Demand (MD) network. Evidence for functional specificity by visual task in these regions has been sparse. We search for such functional specificity here. To this end, we recorded BOLD fMRI responses as human subjects viewed walk-throughs of a virtual 3D map while performing one of three tasks. Participants passively fixated, kept track of their original heading, or kept track of whether one of two target objects had appeared. At the end of each trial, they indicated whether a probe stimulus matched their current heading or the identity of any target. The task structure assured that participants continuously attended to task-relevant information throughout each trial. To dissociate stimulus-driven responses from task-driven responses, the same stimuli occurred in multiple task conditions. We modeled the BOLD responses as a function of either task identity or image-computed features of the stimulus (motion energy and other features). We found that the task model explained unique variance independent of the stimulus-driven models in parietal, frontal, and inferior lateral occipital cortex. We additionally found unique variance explained by task in a region anterior to hMT+ which has not previously been associated with the MD network. Among these regions, we found distinct sets of regions associated with the heading task and the search task. Responses in inferior LOTC were most modulated by the search task, the region near hMT+ by the heading task, and responses along the IPS alternated between the two tasks in a consistent pattern across participants. These results suggest a mosaic of functional specificity for spatial and object-based tasks throughout the MD network.

Patricia Limon, BS

Institution: Stanford University
Talk Title: Effect of Attentional State on Experience Dependent Visual Cortical Neuroplasticity: Contrast Response Function Changes as a Result of Attention to Visual Stimulus Across Days

Abstract

Passive viewing of a flickering checkerboard has been shown to induce potentiation of EEG visual evoked responses in both rodents and humans, providing a noninvasive technique for measuring experience-dependent visual cortical neuroplasticity sometimes referred to as stimulus-specific response potentiation (SSRP). The most robust form of SSRP identified in humans is a 10-minute presentation of a full field high-contrast checkerboard stimulus that contrast-reverses at 2 Hz (See, e.g. Norman et al Biological Psychiatry 2007). It is well known that attention has a powerful effect on perceptual learning and memory, but how attention regulates experience-dependent neuroplasticity, if at all, is little studied and has not been tested in the context of SSRP. Here we decided to test if attentional state affects the neuroplasticity measured with SSRP. We hypothesize that attentional deployment to the induction stimulus enhances response potentiation specifically within the retinotopic extent of the induction stimulus, while attending away from the induction stimulus may suppress response potentiation. This would be evidenced by a larger leftward shift and/or increased slope of the Contrast Response Function when attention is directed to the potentiated stimulus. Our preliminary data (n=27) suggests that post-induction neural activity increased regardless of attentional deployment. Interestingly, when analyzing the potentiated hemifield data across days, our results preliminarily demonstrate a greater SSRP induced response enhancement on the second day compared to the first day. Additionally, response potentiation occurred in both the potentiated (left) and control non-potentiated (right) hemifield, suggesting that SSRP effects can extend beyond the retinotopic extent of the induction stimulus.

Katrin Franke, PhD

Institution: Stanford Unviersity School of Medicine
Talk Title: Mapping neuronal representations in the visual cortex using deep learning
Learn More: Google Scholar

Abstract

My talk will explore how we use deep learning as a powerful tool to enhance our understanding of neuronal representations within the visual cortex. By employing deep neural networks as digital twins for specific cortical areas in both mice and macaques, we facilitate comprehensive in-silico experiments that precede and inform targeted in-vivo verification. This method integrates large-scale neuronal recordings with advanced AI models to systematically analyze how visual neurons encode diverse stimuli attributes, ranging from isolated object features to more complex scenarios involving internal brain states and contextual information. This approach not only deepens our understanding of the functional organization within the visual cortex but also establishes a link between biological and artificial vision systems.

YuYi Lu, BS, BA

Institution: University of California, Berkeley
Talk Title: The locus of redundancy masking

Abstract

In redundancy masking (RM), the perceived number of items in repeating patterns is reduced (e.g., perceiving two lines when three are presented). RM has been found with simple (e.g., lines) and complex stimuli (e.g., letters and faces), suggesting that RM occurs at an early stage or multiple stages of the visual processing hierarchy. Here, we investigated whether the neuronal interactions underlying RM occur at a retinal level or higher up in the visual pathways. Observers were presented with arrays of 3, 4 or 5 vertical lines, radially arranged with respect to central fixation in a mirror stereoscope. The lines were presented either ipsilaterally (all lines displayed to the same or both eyes) or contralaterally (lines displayed to different eyes). The center of the arrays was presented at 10º of eccentricity, either to the left or right of fixation along the horizontal meridian for 150 ms. Center-to-center line spacing was set at 0.78º, which was confirmed to be above observers minimum angle of resolution with a Landolt C orientation discrimination task. Results showed RM (i.e., reporting less lines than presented) for all numbers of lines (3, 4 and 5), with the magnitude of RM increasing with the number of lines presented. Crucially, RM was found both for ipsilateral, as well as contralateral presentations. The current results suggest that the locus of RM is at or beyond the site of binocular integration. These findings are a stepping stone to identifying the precise cortical locus or loci of RM. Future psychophysical and neurophysiological research on RM will reveal whether it occurs solely at early stages (e.g., V1, V2), at later stages (e.g., V3, V4), or whether it is a multistage process with several areas playing a key role in giving rise to this type of masking.

Ryan Ash, MD, PhD

Institution: Stanford University School of Medicine
Talk Title: Noninvasive neuromodulation of human visual thalamus with transcranial ultrasound stimulation
Learn More: Google Scholar

Abstract

Clinical neuroscience has led to a revolution in our understanding of how different neural circuits contribute to phenotypes of neuropsychiatric conditions. Unfortunately, the tools to modulate these circuits in the human brain and thereby ameliorate symptoms are limited by poor focality and depth penetration. Transcranial ultrasound stimulation (TUS) is an emerging tool to achieve noninvasive focal brain-wide neuromodulation with high focality (less than 1cm) and the ability to achieve high intensities in-depth. This technology is at an early stage of development, and many key optimizations are needed to accelerate clinical trials in psychiatry. One key need is to better understand how different sonication parameters relate to neuromodulatory effects. Key parameters include the pulse repetition frequency (PRF), intensity, and duty cycle (DC). Importantly, by varying these parameters it may be possible to have either predominantly suppressive/inhibitory effects or facilitatory/excitatory effects on neural activity and synaptic strength. Previous work by our group and others in large animal models (e.g. Fry et al Science 1958, Mohammadjavadi, Ash et al., Scientific Reports 2022) have demonstrated a suppression of EEG visual-evoked potentials (VEPs) when TUS is targeted to the visual thalamus (lateral geniculate nucleus). We are therefore adapting this paradigm into human, as an efficient testbed to evaluate TUS effects. We implemented steady-state visual evoked potential (ssVEP) measures of contrast-response (Ash, Norcia Psychophysiology 2023) and contrast increment detection psychophysics as robust neural and behavioral readouts of subcortical visual pathway function. We developed a robust neuroimaging and simulation pipeline to target LGN, and we are using a neuronavigated depth-steerable 4-element TUS transducer. Our preliminary data suggests that VEPs can be reversibly suppressed with TUS to the human LGN. We are parametrically varying PRF, intensity, and DC to determine which parameters lead to the most robust suppressive and/or facilitatory effects. This work provides the foundation for a dissection of the roles of subcortical and deep cortical nuclei in cognition, emotion, and sensory processing in health and disease.

Michael Silver, PhD

Institution: University of California, Berkeley
Talk Title: Predictions and visual perceptual selection
Learn More: Website

Abstract

Internally generated predictions about visual events or stimuli can have substantial effects on visual perception. In some cases, there is enhancement of perception for those stimuli that match predictions, while in other cases, perception of predicted stimuli is suppressed to emphasize unexpected stimuli that do not match predictions. I will describe a series of studies in human subjects in which we manipulated the predictability of upcoming visual stimuli and employed binocular rivalry to assess the effects of these predictions on visual perceptual selection. In each binocular rivalry display, one eyes stimulus was consistent with the subjects predictions, while the other eyes stimulus was inconsistent with these predictions. The results from these studies raise the possibility that stimuli that are preferentially processed early in the visual cortical pathways are more likely to be enhanced when they are predicted, while stimuli processed at higher levels of the visual hierarchy are more likely to be enhanced when they are surprising.

Wendy Torrens, MS

Institution: University of Nevada, Reno
Talk Title: Anomalous visual cortical excitability in Schizotypy
Learn More: Google Scholar

Abstract

One of the most robust biomarkers in psychiatry is the sensory hypo-excitability seen in individuals with schizophrenia, especially to auditory stimuli, suggesting that they are less sensitive to incoming sensory information than non-clinical individuals. However, subclinical individuals exhibiting schizophrenia-like traits (schizotypy) report more illusions in the Pattern Glare Test (PGT), consistent with behavioral hyperexcitability and being more sensitive to sensory (visual) input. This throws into question the use of sensory biomarkers to assess symptom load subclinically. We examined cortical visual excitability in schizotypy to assess its use as a biomarker of symptom load. Participants completed the Schizotypal Personality Questionnaire Brief Revised (SPQ-BR) and were categorized into high schizotypy (SPQ greater than 87) and a control group (SPQ-BR = 50-70). Participants completed the PGT, and electroencephalography (EEG) was recorded while they viewed grating patterns that reversed contrast at 6Hz. As expected, individuals with high schizotypy reported more illusions from the PGT than controls (Cohens d = 1.01). They also produced larger visual steady-state responses compared to controls (Cohens d = 0.3). High schizotypy exhibited visual cortical hyperexcitability, consistent with the higher number of illusions seen in the PGT, but inconsistent with what we see in diagnosed schizophrenia. This divergent multisensory profile in schizotypy may be crucial in helping us understand how the underlying pathology changes with symptom progression.

Vaidehi Natu, PhD

Institution: Stanford University
Talk Title: Infants' cortex undergoes microstructural growth coupled with myelination during development
Learn More: Google Scholar

Abstract

The establishment of neural circuitry during early infancy is critical for developing complex sensory and motor functions. However, how cortical tissue develops postnatally is largely unknown. By combining structural MRI and quantitative MRI measure (R1) in infants, we tracked cortical tissue development across 3 time-points (newborn, 3 months, and 6 months). Higher R1 indicates higher microstructural tissue density and more developed cortex. Our data reveal three main findings: First, primary sensory/motor areas (V1: visual, A1: auditory, S1: somatosensory, M1: motor) have higher R1 at birth than higher-level cortical areas. However, all primary areas show significant increase in R1 in first six months of life, illustrating profound tissue growth after birth. Second, significant increase in R1 in newborns to 6-month-olds occurs in all visual areas of the ventral and dorsal visual streams. Strikingly, this development is heterogenous across the visual hierarchies: Earlier areas are more developed with denser tissue at birth than higher-order areas, but higher-order areas have faster rates of development. Finally, analysis of transcriptomic gene data that compares gene expression in postnatal vs. prenatal tissue samples showed strong postnatal expression of genes associated with myelination, synaptic signaling, and dendritic processes. Our results indicate that these cellular processes may contribute to profound postnatal tissue growth in the visual cortex observed in our in-vivo measurements. We propose a novel principle of postnatal maturation of sensory systems: development of cortical tissue proceeds in a hierarchical manner, enabling the lower-level areas to develop first to provide scaffolding for higher-order areas, which begin to develop more rapidly following birth to perform complex computations for sensory-motor processes.

Dan Yamins, PhD

Institution: Stanford University
Talk Title: Climbing the Ladder of Causation Toward Intermediate Vision
Learn More: Google Scholar

Abstract

Humans and (almost certainly) many animals possess extremely rich low-to-mid-level visual scene understanding concepts, including (inter alia) the ability to estimate contours and border ownership, optical flow and self-induced motion, a '2.5D sketch' of monocular depth and surface normals, various kinds of segmentation, 3D shape, and materials. But where do these concepts come from, especially given that real biological organisms are definitely not getting detailed supervision for all these rich visual properties? I will present a working theory of how they arise, based on recent work from my lab on Counterfactual World Models (CWMs). Specifically, I will describe a specific form of masked prediction that enables the training of large-scale predictive models that organically possess high-quality causally-informative tokens. I will then show how a wide variety of mid-level visual concepts arise via performing simple generic interventions on these tokens, and computing counterfactual effects therefrom. The resulting model takes a step towards a generic unsupervised algorithm for visual scene understanding in machines, and helps us formulate novel and intriguing hypotheses for the origins of biological vision.

Jorge Otero-Millan, PhD

Institution: University of California, Berkeley
Talk Title: Learning to hold the eyes still and aligned
Learn More: Website

Abstract

Most common eye movement defects involve either not being able to align the eyes, strabismus, or not being able to hold them still, nystagmus. Superficially, these two symptoms appear unrelated, with strabismus arising from a failure to control eye position to reduce retinal disparity, and nystagmus arising from a failure to control eye velocity to reduce retinal motion. However, many patients suffer from both, highlighting the relevance of mechanisms linking velocity and position control. One potential link appears in the areas of the brain that mathematically convert eye velocity commands into eye position signals. This distributed brain circuit is called the ocular motor integrator (integrator for simplicity). We take advantage of one paradigm of integrator adaptation known as rebound nystagmu (RN). RN occurs in straight ahead gaze after a sustaining eccentric gazing position for about 30 seconds or more. The slow phase or drift of RN is directed toward the previously held eccentric eye position. We will show results of experiments showing how the integrator adapts differently its velocity or position state as well as in horizontal or vertical dimensions. In the future we plan to compare the adaptation to holding eccentric gaze to another phenomenon known as Phoria adaptation that involves adapting to looking at different distances. Understanding how the brain learns to hold the eyes stable and aligned is critical to understanding those different causes, diagnosing them, and developing treatments for them.

Reem Almagati, OD, MSc

Institution: University of California, Berkeley
Talk Title: Effects of Convergence and Monocular viewing on Perception of the World's Horizontal and Vertical in Virtual Reality
Learn More: Student Profile

Abstract

When we converge to look at a near target our eyes also typically undergo incyclotorsion where their top poles rotate towards the nose. In order to perceive orientations in the world accurately the visual system must take into account eye position and not rely only on the orientation of a stimulus on the retina. Here we utilized Virtual Reality (VR) to study how our perception is affected by cyclovergence. Subjects performed Subjective Visual Vertical (SVV) and Subjective Visual Horizontal (SVH) tasks using the FOVE0 VR headset at three head positions: upright, 20 degrees right-ear down and 20 degrees left-ear down. While they fixated on a head-fixed central target, subjects reported whether a line originating from the target, tilted between -12 to 12 degrees, was CW or CCW with respect to the world. The fixation target was presented binocularly while the stimulus line was presented either monocularly or binocularly at two vergence distances: 0.3m (near) or 1.5m (far). Near and far conditions were blocked while the other conditions were randomized. We found the mean difference in perceived vertical (SVV) between the left and right eye viewing conditions across all head positions to be 2.4 ± 0.3 degrees for near and 0.9 ± 0.3 degrees for far. The mean difference in perceived horizontal (SVH) between the left and right eye viewing across all head positions was 1.8 ± 0.4 degrees for near and 0.8 ± 0.3 degrees for far. That is, errors in perceived line orientation showed a CW bias in the left eye viewing condition and a CCW bias in the right eye viewing condition. These results show a bias in the monocular perception of the worlds horizontal and vertical consistent with the expected incyclotorsion at near viewing. Our findings suggest that the visual system does not utilize an extra retinal signal when assessing the orientation of a stimulus relative to the world presented monocularly in a VR setting.

Orhan Soyuhos, MSc

Institution: University of California, Davis
Talk Title: Posterior Parietal Inactivation Alters Intrinsic Neural Timescales and Attentional Processing in Frontal Eye Field Neurons
Learn More: Website

Abstract

Intrinsic neural timescales characterize the dynamics of endogenous fluctuations in neural activity. We measured the intrinsic timescales of frontal eye field (FEF) neurons and examined changes during posterior parietal cortex (PPC) inactivation. FEF neurons exhibit a bimodal distribution of intrinsic timescales, with shorter timescale neurons processing rapid visual information and longer timescale neurons more involved in sustained attentional modulation. PPC inactivation significantly increased intrinsic timescales in both neuron types, with a 15-fold greater increase in shorter timescale neurons. Additionally, PPC inactivation reduced visual and attentional responses, with a stronger effect on attention in longer timescale neurons. This disruption eliminated the correlation between timescales and attentional responses observed in the control condition. Our results provide the first causal evidence that FEF intrinsic timescales depend on long-range projections from PPC, suggesting the presence of at least two network motifs with different timescales that contribute to neuronal dynamics and functional computations within FEF.

Tyson Kim, MD, PhD

Institution: UCSF School of Medicine
Talk Title: Transscleral Multiphoton Imaging and Flowmetry
Learn More: UCSF Profile

Abstract

Optical imaging in the eye traditionally focuses through the pupil to visualize posterior intraocular anatomy. However, many important peripheral and deeper structures involved in blinding conditions are obscured by opaque and pigmented tissues. The Kim lab develops advanced microscopy and analytical methods to image directly through the scleral wall to visualize previously inaccessible regions of the eye with subcellular resolution, enabling new investigations of ocular disease and physiology.

Kiara Abhayaratne

Institution: University of California, Berkeley
Talk Title: Modeling Photoreceptor Loss and Behavioral Performance in Retinitis Pigmentosa in rd10 Mice Using a Cricket-Hunting Assay

Abstract

Retinitis pigmentosa (RP) is a progressive retinal disease that causes vision loss due to the degeneration of rod and cone photoreceptor cells. While the anatomy and morphology of the retina during RP progression has been investigated in previous studies, this study measures the impact of photoreceptor loss on a visually guided behavior in Retinal Degeneration 10 (rd10) mice using a cricket hunting assay. Cricket hunting is an ethologically relevant behavior for assessing vision loss in rodent models of retinal degeneration. The performance of 16 rd10 mice and 8 wild-type (C57BL/6J) mice was measured over time, with age serving as a proxy for the amount of degeneration in rd10 mice. An arena with four mirrors and a camera was used for 3D and 2D tracking and recording of the hunts. DeepLabCut software was employed for markerless tracking of mouse and cricket movements, identifying features of interest (FOI). FOIs included vectored approach-times when the mouse orients its gaze towards and rapidly approaches the cricket, and extended immobilization-times to capture the cricket and consume or hold it for an extended period of time. A custom graphical interface was used to classify these FOIs. We found that rd10 mice showed a progressive increase in time-to-capture crickets, alongside changes in hunting strategies, as their vision deteriorated. These results demonstrate the promise of cricket hunting assays in assessing rodent vision. Future work will focus on evaluating gene therapies for restoring vision in mice with photoreceptor degeneration diseases.

Yujia Yang, MD, PhD

Institution: University of California, Berkeley
Talk Title: Reduced corneal susceptibility to colonization by commensal bacteria after contact lens wear is associated with parainflammation involving Lyz2+ cell trafficking toward the corneal surface
Learn More: Google Scholar

Abstract

Purpose: Contact lens wear in mice can induce corneal para-inflammation similar to human lens wear. We have shown this involves CD11c+, Lyz2+ and gd-T cells after 24 h of wear, and gd-T cells, Lyz2+ cells and Ly6G+ cells after 6-7 days wear. Here, we explored the relationship between the Lyz2+ cell response and corneal susceptibility to colonization by commensal bacteria not normally found on the cornea.
Conclusions: Contact lens wear protects the superficially-injured cornea against the adhesion of commensal bacteria associated with an increased Lyz2+ cell response and their migration towards the corneal surface. The data suggest that contact lens related parainflammation primes the cornea for defense against bacterial colonization correlating with Lyz2+ cell changes. Further studies are needed to show involvement of specific Lyz2+ cell subtypes and to determine the significance of enhanced penetration of adherent bacteria after lens wear.

Return to Main BAVRD Page