Systematic Integration of Structural and Functional Data into Multi-scale Models of Mouse Primary Visual Cortex

Billeh, Yazan N.; Cai, Bin; Gratiy, Sergey L.; Dai, Kael; Iyer, Ramakrishnan; Gouwens, Nathan W.; Abbasi-Asl, Reza; Jia, Xiaoxuan; Siegle, Joshua H.; Olsen, Shawn R.; Koch, Christof; Mihalaş, Ştefan; Arkhipov, Anton

doi:10.1016/j.neuron.2020.01.040

Cited by 243 publications

(651 citation statements)

References 115 publications

(253 reference statements)

Supporting

Mentioning

640

Contrasting

Order By: Relevance

“…While it is desirable to have optimal behavior at both high and low noise levels, it is unclear how complicated the underlying structure must be to realize such a transition. We here demonstrate that our recently published biologically realistic model of mouse V1 (Billeh et al 2020) precisely shows such a transition without requiring any alteration or parameter tuning. This shows the power of realistic models to generalize, and link to theoretical aspects outside of those for which they were trained.…”

Section: Introductionmentioning

confidence: 59%

“…The GLIF V1 model, detailed in (Billeh et al 2020), is visualized in Fig. 1A-C, and briefly outlined below.…”

Section: Resultsmentioning

confidence: 99%

“…Since a comprehensive experimental characterization of the input/output transfer functions is out of reach, we seek to perform such a characterization using a comprehensive model of one cortical area. We have constructed such a model for the mouse primary visual cortex (area V1) at two levels of neuronal granularity, generalized leaky integrate-and-fire neurons (GLIF) and biophysically-detailed neurons with spatially extended dendritic trees (Billeh et al 2020). The model is thoroughly constrained by experimental data in terms of distribution of cell types (Teeter et al 2018;Gouwens et al 2018), connectivity and visual inputs (Durand et al 2016), and reproduces a variety of in vitro and in vivo observations of cellular activity under both twophoton calcium imaging as well as Neuropixels recordings (Siegle et al 2019;de Vries et al 2019).…”

Section: Introductionmentioning

confidence: 99%

“…In this study, we used this newly-built V1 model employing the GLIF neuron (Teeter et al 2018) representation to simulate experiments. It requires three orders of magnitude less computing time than the biophysically detailed version (Billeh et al 2020;Gouwens et al 2018), allowing us to more rapidly explore the space of inputs by conducting thousands of simulations, permitting exploration of the characteristics and coding mechanisms of V1 in an efficient way.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Modeling robust and efficient coding in the mouse primary visual cortex using computational perturbations

Cai

Billeh

Chettih

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Investigating how visual inputs are encoded in visual cortex is important for elucidating the roles of cell populations in circuit computations. We here use a recently developed, large-scale model of mouse primary visual cortex (V1) and perturb both single neurons as well as functional-and cell-type defined population of neurons to mimic equivalent optogenetic perturbations. First, perturbations were performed to study the functional roles of layer 2/3 excitatory neurons in inter-laminar interactions. We observed activity changes consistent with the canonical cortical model (Douglas and Martin 1991). Second, single neuron perturbations in layer 2/3 revealed a center-surround inhibition-dominated effect, consistent with recent experiments. Finally, perturbations of multiple excitatory layer 2/3 neurons during visual stimuli of varying contrasts indicated that the V1 model has both efficient and robust coding features. The circuit transitions from predominantly broad like-to-like inhibition at high contrasts to predominantly specific liketo-like excitation at low contrasts. These in silico results demonstrate how the circuit can shift from redundancy reduction to robust codes as a function of stimulus contrast.

show abstract

Section: Introductionmentioning

confidence: 59%

“…The GLIF V1 model, detailed in (Billeh et al 2020), is visualized in Fig. 1A-C, and briefly outlined below.…”

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Modeling robust and efficient coding in the mouse primary visual cortex using computational perturbations

Cai

Billeh

Chettih

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…3A, B). We first characterized direction selectivity as a function of the TF and SF of the drifting grating stimulus using mouse parameters 18,19 . We find that the relative phase of the subfield responses can shift from completely out-of-phase to completely in-phase depending on both the direction and SF of the stimulus (Fig.…”

mentioning

confidence: 99%

Widespread presence of direction-reversing neurons in the mouse visual system

Billeh

Iyer

Wahle

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

Direction selectivity -the preference of motion in one direction over the opposite -is a fundamental property of visual neurons across species. We find that a substantial proportion of direction selective neurons in the mouse visual system reverse their preferred direction of motion in response to drifting gratings at different spatiotemporal parameters. A spatiotemporally asymmetric filter model recapitulates our experimental observations.Motion detection is a feature common to all visual animals, and recent work has shown strong parallels in these computations across many species 1-9 . Similar circuits in these animals underlie the comparison of visual information at two locations at two points in time, a computation that establishes direction selectivity wherein a neuron preferentially responds to motion in one direction over its opposite, or null, direction. Here, we report neurons in the mouse visual system that respond preferentially to motion in their null direction at different spatial and temporal frequencies.Using data from the Allen Brain Observatory, a large-scale survey of visual responses in the mouse visual cortex recorded using 2-photon (2P) calcium imaging 10 , we analyzed responses to the drifting grating stimulus. This stimulus consisted of full-field sinusoidal gratings moving in 8 directions and at 5 temporal frequencies (TFs), but at a single spatial frequency (SF). Many neurons showed direction selectivity, and among these direction selective neurons we found some that reverse their direction preference in response to gratings at different TFs (Fig. 1A). We use the direction selectivity index (DSI, see Methods) to quantify the strength of direction selectivity at each TF, fixing the preferred and null directions to those determined at the preferred TF (TF pref , the TF evoking the largest mean response). A negative DSI thus indicates a reversal of direction preference (Fig. 1B, C). We term the neurons exhibiting this phenomenon Direction Reversing Neurons (DRNs).We imposed strict criteria to identify DRNs by comparing their responses at their preferred condition (the preferred direction at TF pref , blue box in Fig. 1A) and reversed condition (the TF at which the neuron has its largest mean response in the null direction, orange box in Fig. 1A). Candidate DRNs must have a DSI ≥ 0.3 for the preferred condition and exhibit a larger response to the null than preferred direction at the reversed condition. These neurons must also pass a strict bootstrap test (see Methods) to ensure the reversal is not a chance observation driven by a small number of outlier trials. 708 out of 12,515 direction selective neurons (6%) met these criteria. Furthermore, we observe DRNs in all visual areas recorded ( Fig. 1D) and across all transgenically defined populations available in the dataset (Fig. SI1).

show abstract

ORCANet: Differentiable multi‐parameter learning for crowd simulation

Zhang

Wang

et al. 2022

Computer Animation & Virtual

View full text Add to dashboard Cite

Realistic crowd simulation has always been an important research field in computer graphics. While both agent-based motion models and data-driven behavior models have made some progress, they are still suffering from either huge effort of multi-parameter tuning or limited realistic motion. In this article, we propose a novel and differentiable multi-parameter learning method for crowd simulation, which is called ORCANet. The main idea is to learn from real data and inverse evaluating the multi-parameter for subsequent simulation.ORCANet uses classic optimal reciprocal collision avoidance (ORCA) as a basic motion model which is integrated into the deep learning framework. Addressing the feature of linear programming and non-differentiable operation, a Gaussian kernel is added to approximate the role of neighbor distance in collision avoidance, which turns the original discrete operation into a fully differentiable forward simulation. Furthermore, we leverage ORCANet to optimize the multi-parameter combination in synthetic and real-world datasets. ORCANet is proved to rapidly converge to correct parameter values and regenerate the input synthetic sequence. Moreover, experiments on real-world datasets by the metric of pedestrian trajectories verified that a more realistic crowd simulation has been generated through ORCANet.

show abstract

Systematic Integration of Structural and Functional Data into Multi-scale Models of Mouse Primary Visual Cortex

Cited by 243 publications

References 115 publications

Modeling robust and efficient coding in the mouse primary visual cortex using computational perturbations

Modeling robust and efficient coding in the mouse primary visual cortex using computational perturbations

Widespread presence of direction-reversing neurons in the mouse visual system

ORCANet: Differentiable multi‐parameter learning for crowd simulation

Contact Info

Product

Resources

About