Neuroscience and Neurobiology Commons^™

Validity Of Neural Distance Measures In Representational Similarity Analysis, Fabian A. Soto, Emily R. Martin, Hyeonjeong Lee, Nafiz Ahmed, Juan Estepa, Kianoosh Hosseini, Olivia A. Stibolt, Valentina Roldan, Alycia Winters, Mohammadreza Bayat

MODVIS Workshop

No abstract provided.

Visual Expertise In An Anatomically-Inspired Model Of The Visual System, Garrison W. Cottrell, Martha Gahl, Shubham Kulkarni

MODVIS Workshop

We report on preliminary results of an anatomically-inspired deep learning model of the visual system and its role in explaining the face inversion effect. Contrary to the generally accepted wisdom, our hypothesis is that the face inversion effect can be accounted for by the representation in V1 combined with the reliance on the configuration of features due to face expertise. We take two features of the primate visual system into account: 1) The foveated retina; and 2) The log-polar mapping from retina to V1. We simulate acquisition of faces, etc., by gradually increasing the number of identities the network learns. …

Characterization Of Local And Global Statistics In Three Kinds Of Medical Images, And An Example Of Their Role In A Clinical Judgment, Jonathan Victor, Amanda Simon, Craig K. Abbey

MODVIS Workshop

No abstract provided.

A Two-Layer Model Explains Higher-Order Feature Selectivity Of V2 Neurons, Timothy D. Oleskiw, Justin D. Lieber, J. Anthony Movshon, Eero P. Simoncelli

MODVIS Workshop

Neurons in cortical area V2 respond selectively to higher-order visual features, such as the quasi-periodic structure of natural texture. However, a functional account of how V2 neurons build selectivity for complex natural image features from their inputs – V1 neurons locally tuned for orientation and spatial frequency – remains elusive.

We made single-unit recordings in area V2 in two fixating rhesus macaques. We presented stimuli composed of multiple superimposed grating patches that localize contrast energy in space, orientation, and scale. V2 activity is modeled via a two-layer linear-nonlinear network, optimized to use a sparse combination of V1-like outputs to account …

A Bayesian Account Of Depth From Shadow, James Elder, Patrick Cavanagh, Roberto Casati

MODVIS Workshop

When an object casts a shadow on a background surface, the offset of the shadow can be a compelling cue to the relative depth between the object and the background (e.g., Kersten et al 1996, Fig. 1). Cavanagh et al (2021) found that, at least for small shadow offsets, perceived depth scales almost linearly with shadow offset. Here we ask whether this finding can be understood quantitatively in terms of Bayesian decision theory.

Estimating relative depth from shadow offset is complicated by the fact that the shadow offset is co-determined by the slant of the light source relative to the …

Fixational Eye Movements, Perceptual Filling-In, And Perceptual Fading Of Grayscale Images, Michael E. Rudd

MODVIS Workshop

No abstract provided.

Constraining Computational Models Of Brightness Perception: What’S The Right Psychophysical Data?, Guillermo Aguilar, Joris Vincent, Marianne Maertens

MODVIS Workshop

No abstract provided.

Identifying And Localizing Multiple Objects Using Artificial Ventral And Dorsal Visual Cortical Pathways, Zhixian Han, Anne Sereno

MODVIS Workshop

We concluded in our previous study that model cortical visual pathways actively retained information differently according to the different goals of the training tasks. One limitation of our study was that there was only one object in each input image whereas in reality there may be multiple objects in a scene. In our current study, we try to find a brain-like algorithm that can recognize and localize multiple objects.

Model Of Visual Contrast Gain Control And Pattern And Noise Masking, Joshua A. Solomon

MODVIS Workshop

The first stage of the model can be subdivided into a global contrast sensitivity function (a 2-D log-parabolic filter of spatial frequency), followed by an array of sensors having Gabor-pattern receptive fields. The second stage is contrast gain control. At this stage, sensor outputs are subjected to an expansive transformation. Then the outputs are pooled and used to inhibit (or “normalize”) each other. Inhibition is strongest between sensors with similar preferences for orientation, spatial frequency and spatial location. In the final stage of the model, the nomalized sensor outputs for each image are subjected to Minkowski pooling. Two-alternative, forced-choice detection …

Is The Selective Tuning Model Of Visual Attention Still Relevant?, John K. Tsotsos

MODVIS Workshop

No abstract provided.

Functional Organization Of Cortical Maps For Ocular Dominance And Light-Dark Polarity In Primary Visual Cortex, Sohrab Najafian, Jian Zhong Jin, Jose-Manuel Alonso

MODVIS Workshop

No abstract provided.

Computations Of Top-Down Attention By Modulating V1 Dynamics, David Berga, Xavier Otazu

MODVIS Workshop

The human visual system processes information defining what is visually conspicuous (saliency) to our perception, guiding eye movements towards certain objects depending on scene context and its feature characteristics. However, attention has been known to be biased by top-down influences (relevance), which define voluntary eye movements driven by goal-directed behavior and memory. We propose a unified model of the visual cortex able to predict, among other effects, top-down visual attention and saccadic eye movements. First, we simulate activations of early mechanisms of the visual system (RGC/LGN), by processing distinct image chromatic opponencies with Gabor-like filters. Second, we use a cortical …

Differentiating Changes In Population Encoding Models With Psychophysics And Neuroimaging, Jason Hays, Fabian Soto Phd

MODVIS Workshop

It is now common among visual scientists to make inferences about neural population coding of stimuli from indirect measures such as those provided by neuroimaging and psychophysics. The success of such studies depends strongly on simulation work using standard population encoding models extended with decoders (in psychophysics) and measurement models (in neuroimaging). However, not all studies are accompanied by simulation work, and those that are tend to vary widely in their assumptions about encoding, decoding, and measurement. To solve these issues, we designed a Python package (PEMGUIN) to assist computational modelling by providing simple ways to manage encoders' tuning functions, …

Virtual Eye: A Spatial-Temporal Bottom-Up Eye Sensitivity Model, Todd Goodall

MODVIS Workshop

Video quality and compression models use the

spatial contrast sensitivity function (CSF), which is solved

based on a linear system approximation. This function measures

the eye’s sensitivity to sinusoid gratings, ignoring the subtle

connectivity and inhomogeniety of cell density across the

visual field. Non-linear aspects of the eye, such as the change

in frequency sensitivity with changing illumination, are not

captured by this simple approximation. We propose Virtual

Eye, a bottom-up approach that models the spatio-temporal

dynamics of the eye across the visual field. Each functional

retinal cell layer in the eye is modeled using non-uniform spatial

cell responses, which …

Variance Partitioning Reveals Consistent Representation Of Object Boundary Contours In Lo Across Different Datasets, Mark D. Lescroart, Utkarsh Singhal

MODVIS Workshop

No abstract provided.

The Challenge For Vision Of Fluctuating Real-World Illumination, David H. Foster

MODVIS Workshop

No abstract provided.

‘Preferred’ Stimulus Of A Whole Model Visual System, Olivier Penacchio, Julie M. Harris

MODVIS Workshop

No abstract provided.

Finding Any Waldo: Zero-Shot Invariant And Efficient Visual Search, Gabriel Kreiman, Mengmi Zhang

MODVIS Workshop

Visual search constitutes a ubiquitous challenge in natural vision, including daily tasks such as finding a friend in a crowd or searching for a car in a parking lot. Visual search must fulfill four key properties: selectivity (to distinguish the target from distractors in a cluttered scene), invariance (to localize the target despite changes in its rotation, scale, illumination, and even searching for generic object categories), speed (to efficiently localize the target without exhaustive sampling), and generalization (to search for any object, even ones that we have had minimal or no experience with). Here we propose a computational model that …

Linking Signal Detection Theory And Encoding Models To Reveal Independent Neural Representations From Neuroimaging Data, Fabian A. Soto

MODVIS Workshop

No abstract provided.

Texture Statistics Are Sufficient For Ensemble Perception, Sasen S. Cain, Matthew S. Cain

MODVIS Workshop

No abstract provided.

Michelson Contrast For Transparency Perception In Scenes With Multiple Luminances, Minjung Kim, Guillermo Aguilar, Marianne Maertens

MODVIS Workshop

No abstract provided.

Modeling Neural Computations In Lgn And Visual Cortex That Underlie Contextual Modulation Of Lightness And Darkness Magnitudes In Simple And Complex Images, Michael E. Rudd

MODVIS Workshop

No abstract provided.

Brightness Perception Involves Local Adaptation Opposed By Lateral Interaction, Qasim Zaidi, Romain Bachy, Jose-Manuel Alonso

MODVIS Workshop

No abstract provided.

The Road Towards Image-Computable Models Of Human Visual Grasp Planning, Guido Maiello, Lina K. Klein, Vivian C. Paulun, Katherine R. Storrs, Roland W. Fleming

MODVIS Workshop

No abstract provided.

A Feature-Based Model Of Visually Perceiving Deformable Objects, Vivian C. Paulun, Filipp Schmidt, Roland W. Fleming

MODVIS Workshop

No abstract provided.

Global Estimation Of Signed 3d Surface Tilt From Natural Images, Seha Kim, Johannes Burge

MODVIS Workshop

The ability of human visual systems to estimate 3D surface orientation from 2D retinal images is critical. But the computation to calculate 3D orientation in real-world scenes is not fully understood. A Bayes optimal model grounded in natural statistics has explained 3D surface tilt estimation of human observers in natural scenes (Kim and Burge, 2018). However, the model is limited because it estimates only unsigned tilt (tilt modulo 180deg). We extend the model to predict signed tilt estimates and compared with human signed estimates. The model takes image pixels as input and produces optimal estimates of tilt as output, using …

Inferring The Neural Representation Of Faces From Adaptation Aftereffects, Kara J. Emery, Michael A. Webster Ph.D.

MODVIS Workshop

The aftereffects of adaptation to faces have been studied widely, in part to characterize the coding schemes for representing different facial attributes. Often these aftereffects have been interpreted in terms of two alternative models of face processing: 1) a norm-based or opponent code, in which the facial dimension is represented by relative activity in a pair of broadly-tuned mechanisms with opposing sensitivities; or 2) an exemplar code, in which the dimension is sampled by multiple channels narrowly-tuned to different levels of the stimulus. Evidence for or against these alternatives is based on the different patterns of aftereffects they predict (e.g. …

Effect Of Noise On Mutually Inhibiting Pyramidal Cells In Visual Cortex: Foundation Of Stochasticity In Bi-Stable Perception, Naoki Kogo, Felix Kern, Thomas Nowotny, Raymond Van Ee, Richard Van Wezel, Takeshi Aihara

MODVIS Workshop

Bi-stable perception has been an important tool to investigate how visual input is interpreted and how it reaches consciousness. To explain the mechanisms of this phenomenon, it has been assumed that a mutual inhibition circuit plays a key role. It is possible that this circuit functions to resolve ambiguity of input image by quickly shifting the balance of competing signals in response to conflicting features. Recently we established an in vitro neural recording system combined with computerized connections mediated by model neurons and synapses (“dynamic clamp” system). With this system, mutual inhibition circuit between two pyramidal cells from primary visual …

Visual Category Learning By Means Of Basal Ganglia, Fred H. Hamker, Francesc Villagrasa, Javier Baladron, Henning Schroll, Julien Vitay

MODVIS Workshop

No abstract provided.

Why Latent Representations In Convolutional Neural Networks Fall Outside Visual Space, Katerina Malakhova

MODVIS Workshop

It is common to compare properties of visual information processing by artificial neural networks and the primate visual system.

Some remarkable similarities were observed in the responses of neurons in IT cortex and units in higher layers of CNNs. Here I show that latent representations formed by weights in convolutional layers do not necessarily reflect visual domain. Instead they are strongly dependent on a choice of training set and cost function.

The most striking example is when an individual unit, which is highly selective to some members of a category is, nevertheless, inhibited by visually similar objects of the same …