scholarly journals A Convolutional Network Architecture Driven by Mouse Neuroanatomical Data

2020 ◽  
Author(s):  
Jianghong Shi ◽  
Michael A. Buice ◽  
Eric Shea-Brown ◽  
Stefan Mihalas ◽  
Bryan Tripp
Keyword(s):  
Visual Cortex ◽  
Visual System ◽  
Network Architecture ◽  
Structural Parameters ◽  
Cortical Layer ◽  
Hierarchical Organization ◽  
Visually Guided ◽  
Convolutional Network ◽  
Mouse Visual Cortex ◽  
Neuroanatomical Data

Convolutional neural networks trained on object recognition derive some inspiration from the neuroscience of the visual system in primates, and have been used as models of the feedforward computation performed in the primate ventral stream. In contrast to the hierarchical organization of primates, the visual system of the mouse has flatter hierarchy. Since mice are capable of visually guided behavior, this raises questions about the role of architecture in neural computation. In this work, we introduce a framework for building a biologically constrained convolutional neural network model of lateral areas of the mouse visual cortex. The structural parameters of the network are derived from experimental measurements, specifically estimates of numbers of neurons in each area and cortical layer, the interareal connec-tome, and the statistics of connections between cortical layers. This network is constructed to support detailed task-optimized models of mouse visual cortex, with neural populations that can be compared to specific corresponding populations in the mouse brain. The code is freely available to support such research.

scholarly journals CNN MouseNet: A biologically constrained convolutional neural network model for mouse visual cortex

2021 ◽  
Author(s):  
Jianghong Shi ◽  
Bryan Tripp ◽  
Eric Shea-Brown ◽  
Stefan Mihalas ◽  
Michael Buice
Keyword(s):  
Neural Network ◽  
Visual Cortex ◽  
Visual System ◽  
Convolutional Neural Network ◽  
Mouse Brain ◽  
Network Model ◽  
Neural Network Model ◽  
Structural Parameters ◽  
Neural Representation ◽  
Mouse Visual Cortex

Convolutional neural networks trained on object recognition derive inspiration from the neural architecture of the visual system in primates, and have been used as models of the feedforward computation performed in the primate ventral stream. In contrast to the deep hierarchical organization of primates, the visual system of the mouse has a shallower arrangement. Since mice and primates are both capable of visually guided behavior, this raises questions about the role of architecture in neural computation. In this work, we introduce a novel framework for building a biologically constrained convolutional neural network model of the mouse visual cortex. The architecture and structural parameters of the network are derived from experimental measurements, specifically the 100-micrometer resolution interareal connectome, the estimates of numbers of neurons in each area and cortical layer, and the statistics of connections between cortical layers. This network is constructed to support detailed task-optimized models of mouse visual cortex, with neural populations that can be compared to specific corresponding populations in the mouse brain. Using a well-studied image classification task as our working example, we demonstrate the computational capability of this mouse-sized network. Given its relatively small size, MouseNet achieves roughly 2/3rds the performance level on ImageNet as VGG16. In combination with the large scale Allen Brain Observatory Visual Coding dataset, we use representational similarity analysis to quantify the extent to which MouseNet recapitulates the neural representation in mouse visual cortex. Importantly, we provide evidence that optimizing for task performance does not improve similarity to the corresponding biological system beyond a certain point. We demonstrate that the distributions of some physiological quantities are closer to the observed distributions in the mouse brain after task training. We encourage the use of the MouseNet architecture by making the code freely available.

scholarly journals Neuronal adaptation reveals a suboptimal decoding of orientation tuned populations in the mouse visual cortex

2018 ◽  
Author(s):  
Miaomiao Jin ◽  
Jeffrey M. Beck ◽  
Lindsey L. Glickfeld
Keyword(s):  
Visual Cortex ◽  
Visual System ◽  
Cortical Neurons ◽  
Signal To Noise Ratio ◽  
Sensory Information ◽  
Orientation Discrimination ◽  
Signal To Noise ◽  
Neuronal Adaptation ◽  
Related Information ◽  
Mouse Visual Cortex

AbstractSensory information is encoded by populations of cortical neurons. Yet, it is unknown how this information is used for even simple perceptual choices such as discriminating orientation. To determine the computation underlying this perceptual choice, we took advantage of the robust adaptation in the mouse visual system. We find that adaptation increases animals’ thresholds for orientation discrimination. This was unexpected since optimal computations that take advantage of all available sensory information predict that the shift in tuning and increase in signal-to-noise ratio in the adapted condition should improve discrimination. Instead, we find that the effects of adaptation on behavior can be explained by the appropriate reliance of the perceptual choice circuits on target preferring neurons, but the failure to discount neurons that prefer the distractor. This suggests that to solve this task the circuit has adopted a suboptimal strategy that discards important task-related information to implement a feed-forward visual computation.

scholarly journals Unsupervised Models of Mouse Visual Cortex

2021 ◽  
Author(s):  
Aran Nayebi ◽  
Nathan C. L. Kong ◽  
Chengxu Zhuang ◽  
Justin L. Gardner ◽  
Anthony M. Norcia ◽  
...  
Keyword(s):  
Visual Cortex ◽  
Visual System ◽  
Network Architectures ◽  
Deep Convolutional Neural Networks ◽  
Visual Stream ◽  
Computational Theory ◽  
Visual Coding ◽  
Ventral Visual Stream ◽  
Mouse Visual Cortex ◽  
Image Transformations

Task-optimized deep convolutional neural networks are the most quantitatively accurate models of the primate ventral visual stream. However, such networks are implausible as a model of the mouse visual system because mouse visual cortex has a known shallower hierarchy and the supervised objectives these networks are typically trained with are likely neither ethologically relevant in content nor in quantity. Here we develop shallow network architectures that are more consistent with anatomical and physiological studies of mouse visual cortex than current models. We demonstrate that hierarchically shallow architectures trained using contrastive objective functions applied to visual-acuity-adapted images achieve neural prediction performance that exceed those of the same architectures trained in a supervised manner and result in the most quantitatively accurate models of the mouse visual system. Moreover, these models' neural predictivity significantly surpasses those of supervised, deep architectures that are known to correspond well to the primate ventral visual stream. Finally, we derive a novel measure of inter-animal consistency, and show that the best models closely match this quantity across visual areas. Taken together, our results suggest that contrastive objectives operating on shallow architectures with ethologically-motivated image transformations may be a biologically-plausible computational theory of visual coding in mice.

scholarly journals A Sparse Unreliable Distributed Code Underlies the Limits of Behavioral Discrimination

2018 ◽  
Author(s):  
Balaji Sriram ◽  
Alberto Cruz-Martin ◽  
Lillian Li ◽  
Pamela Reinagel ◽  
Anirvan Ghosh
Keyword(s):  
Visual Cortex ◽  
Visual System ◽  
Single Unit ◽  
Discrimination Task ◽  
Orientation Discrimination ◽  
Behavioral Performance ◽  
High Contrast ◽  
The World ◽  
Mouse Visual Cortex ◽  
Linear Decoding

ABSTRACTThe cortical code that underlies perception must enable subjects to perceive the world at timescales relevant for behavior. We find that mice can integrate visual stimuli very quickly (<100 ms) to reach plateau performance in an orientation discrimination task. To define features of cortical activity that underlie performance at these timescales, we measured single unit responses in the mouse visual cortex at timescales relevant to this task. In contrast to high contrast stimuli of longer duration, which elicit reliable activity in individual neurons, stimuli at the threshold of perception elicit extremely sparse and unreliable responses in V1 such that the activity of individual neurons do not reliably report orientation. Integrating information across neurons, however, quickly improves performance. Using a linear decoding model, we estimate that integrating information over 50-100 neurons is sufficient to account for behavioral performance. Thus, at the limits of perception the visual system is able to integrate information across a relatively small number of highly unreliable single units to generate reliable behavior.

scholarly journals Orientation tuning depends on spatial frequency in mouse visual cortex

2016 ◽  
Author(s):  
Inbal Ayzenshtat ◽  
Jesse Jackson ◽  
Rafael Yuste
Keyword(s):  
Visual Cortex ◽  
Spatial Frequency ◽  
Visual System ◽  
Primary Visual Cortex ◽  
Receptive Fields ◽  
Orientation Selectivity ◽  
Natural Scenes ◽  
Response Properties ◽  
Layer 2 ◽  
Mouse Visual Cortex

AbstractThe response properties of neurons to sensory stimuli have been used to identify their receptive fields and functionally map sensory systems. In primary visual cortex, most neurons are selective to a particular orientation and spatial frequency of the visual stimulus. Using two-photon calcium imaging of neuronal populations from the primary visual cortex of mice, we have characterized the response properties of neurons to various orientations and spatial frequencies. Surprisingly, we found that the orientation selectivity of neurons actually depends on the spatial frequency of the stimulus. This dependence can be easily explained if one assumed spatially asymmetric Gabor-type receptive fields. We propose that receptive fields of neurons in layer 2/3 of visual cortex are indeed spatially asymmetric, and that this asymmetry could be used effectively by the visual system to encode natural scenes.Significance StatementIn this manuscript we demonstrate that the orientation selectivity of neurons in primary visual cortex of mouse is highly dependent on the stimulus SF. This dependence is realized quantitatively in a decrease in the selectivity strength of cells in non-optimum SF, and more importantly, it is also evident qualitatively in a shift in the preferred orientation of cells in non-optimum SF. We show that a receptive-field model of a 2D asymmetric Gabor, rather than a symmetric one, can explain this surprising observation. Therefore, we propose that the receptive fields of neurons in layer 2/3 of mouse visual cortex are spatially asymmetric and this asymmetry could be used effectively by the visual system to encode natural scenes.Highlights–Orientation selectivity is dependent on spatial frequency.–Asymmetric Gabor model can explain this dependence.

scholarly journals Canonical and noncanonical features of the mouse visual cortical hierarchy

2020 ◽  
Author(s):  
Rinaldo D. D’Souza ◽  
Quanxin Wang ◽  
Weiqing Ji ◽  
Andrew M. Meier ◽  
Henry Kennedy ◽  
...  
Keyword(s):  
Visual Cortex ◽  
Receptive Field ◽  
Active Vision ◽  
Structural Features ◽  
Hierarchical Organization ◽  
Hierarchical Level ◽  
Continuous Measure ◽  
Axonal Projections ◽  
Visual Cortical ◽  
Mouse Visual Cortex

ABSTRACTNeocortical circuit computations underlying active vision are performed by a distributed network of reciprocally connected, functionally specialized areas. Mouse visual cortex is a dense, hierarchically organized network, comprising subnetworks that form preferentially interconnected processing streams. To determine the detailed layout of the mouse visual hierarchy, laminar patterns formed by interareal axonal projections, originating in each of ten visual areas were analyzed. Reciprocally connected pairs of areas, and shared targets of pairs of source areas, exhibited structural features consistent with a hierarchical organization. Beta regression analyses, which estimated a continuous measure of hierarchical distance, indicated that the network comprises multiple hierarchies embedded within overlapping processing levels. Single unit recordings showed that within each processing stream, receptive field sizes typically increased with increasing hierarchical level; however, ventral stream areas showed overall larger receptive field diameters. Together, the results reveal canonical and noncanonical hierarchical network motifs in mouse visual cortex.

scholarly journals The influence of cortical depth on neuronal responses in mouse visual cortex

2018 ◽  
Author(s):  
Philip O’Herron ◽  
John Woodward ◽  
Prakash Kara
Keyword(s):  
Visual Cortex ◽  
Neuronal Response ◽  
Superficial Layer ◽  
Cortical Layer ◽  
Contrast Imaging ◽  
Response Properties ◽  
Layer 2 ◽  
Layer 4 ◽  
Thalamic Projections ◽  
Mouse Visual Cortex

AbstractWith the advent of two-photon imaging as a tool for systems neuroscience, the mouse has become a preeminent model system for studying sensory processing. One notable difference that has been found however, between mice and traditional model species like cats and primates is the responsiveness of the cortex. In the primary visual cortex of cats and primates, nearly all neurons respond to simple visual stimuli like drifting gratings. In contrast, imaging studies in mice consistently find that only around half of the neurons respond to such stimuli. Here we show that visual responsiveness is strongly dependent on the cortical depth of neurons. Moving from superficial layer 2 down to layer 4, the percentage of responsive neurons increases dramatically, ultimately reaching levels similar to what is seen in other species. Over this span of cortical depth, neuronal response amplitude also increases and orientation selectivity moderately decreases. These depth dependent response properties may be explained by the distribution of thalamic inputs in mouse V1. Unlike in cats and primates where thalamic projections to the granular layer are constrained to layer 4, in mice they spread up into layer 2/3, qualitatively matching the distribution of response properties we see. These results show that the analysis of neural response properties must take into consideration not only the overall cortical lamina boundaries but also the depth of recorded neurons within each cortical layer. Furthermore, the inability to drive the majority of neurons in superficial layer 2/3 of mouse V1 with grating stimuli indicates that there may be fundamental differences in the role of V1 between rodents and other mammals.

scholarly journals The functional specialization of visual cortex emerges from training parallel pathways with self-supervised predictive learning

2021 ◽  
Author(s):  
Shahab Bakhtiari ◽  
Patrick Mineault ◽  
Timothy Lillicrap ◽  
Christopher Pack ◽  
Blake Richards
Keyword(s):  
Visual Cortex ◽  
Loss Function ◽  
Network Architecture ◽  
Visual Recognition ◽  
Functional Specialization ◽  
Neural Network Architecture ◽  
Parallel Pathways ◽  
Predictive Learning ◽  
Using Data ◽  
Mouse Visual Cortex

The visual system of mammals is comprised of parallel, hierarchical specialized pathways. Different pathways are specialized in so far as they use representations that are more suitable for supporting specific downstream behaviours. In particular, the clearest example is the specialization of the ventral ('what') and dorsal ('where') pathways of the visual cortex. These two pathways support behaviours related to visual recognition and movement, respectively. To-date, deep neural networks have mostly been used as models of the ventral, recognition pathway. However, it is unknown whether both pathways can be modelled with a single deep ANN. Here, we ask whether a single model with a single loss function can capture the properties of both the ventral and the dorsal pathways. We explore this question using data from mice, who like other mammals, have specialized pathways that appear to support recognition and movement behaviours. We show that when we train a deep neural network architecture with two parallel pathways using a self-supervised predictive loss function, we can outperform other models in fitting mouse visual cortex. Moreover, we can model both the dorsal and ventral pathways. These results demonstrate that a self-supervised predictive learning approach applied to parallel pathway architectures can account for some of the functional specialization seen in mammalian visual systems.

scholarly journals Brain-Inspired Algorithms for Processing of Visual Data

2021 ◽  
pp. 105-115
Author(s):  
Nicola Strisciuglio ◽  
Nicolai Petkov
Keyword(s):  
Image Processing ◽  
Computer Vision ◽  
Visual Cortex ◽  
Visual System ◽  
Computational Models ◽  
Hierarchical Organization ◽  
Visual Data ◽  
Convolutional Networks ◽  
Improved Stability ◽  
The Brain

AbstractThe study of the visual system of the brain has attracted the attention and interest of many neuro-scientists, that derived computational models of some types of neuron that compose it. These findings inspired researchers in image processing and computer vision to deploy such models to solve problems of visual data processing.In this paper, we review approaches for image processing and computer vision, the design of which is based on neuro-scientific findings about the functions of some neurons in the visual cortex. Furthermore, we analyze the connection between the hierarchical organization of the visual system of the brain and the structure of Convolutional Networks (ConvNets). We pay particular attention to the mechanisms of inhibition of the responses of some neurons, which provide the visual system with improved stability to changing input stimuli, and discuss their implementation in image processing operators and in ConvNets.

Sign in / Sign up

Export Citation Format

Share Document