Model Constrained by Visual Hierarchy Improves Prediction of Neural Responses to Natural Scenes

Antolík, Ján; Hofer, Sonja B.; Bednar, James A.; Mrsic-Flogel, Thomas D.

doi:10.1371/journal.pcbi.1004927

Cited by 55 publications

(94 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further experiments could be done to verify this recurrent computation prediction. If feasible, imaging a population of MT neurons and fitting a population-level model could reveal these recurrent computations, as has been done in V1 (Cossell et al, 2015;Antolík et al, 2016;Klindt et al, 2017). Examining dynamics of tuning to compound stimuli, possibly with whole-cell recording techniques, could also provide empirical evidence regarding the nature of suppression in…”

Section: Conclusion Drawn From the Separable Modelmentioning

confidence: 99%

Compound stimuli reveal the structure of visual motion selectivity in macaque MT neurons

Zaharia

Goris

Movshon

et al. 2019

Preprint

View full text Add to dashboard Cite

Motion selectivity in primary visual cortex (V1) is approximately separable in orientation, spatial frequency, and temporal frequency ("frequency-separable"). Models for area MT neurons posit that their selectivity arises by combining direction-selective V1 afferents whose tuning is organized around a tilted plane in the frequency domain, specifying a particular direction and speed ("velocity-separable"). This construction explains "pattern direction selective" MT neurons, which are velocity-selective but relatively invariant to spatial structure, including spatial frequency, texture and shape. Surprisingly, when tested with single drifting gratings, most MT neurons' responses are fit equally well by models with either form of separability. However, responses to plaids (sums of two moving gratings) tend to be better described as velocity-separable, especially for pattern neurons. We conclude that direction selectivity in MT is primarily computed by summing V1 afferents, but pattern-invariant velocity tuning for complex stimuli may arise from local, recurrent interactions. Significance StatementHow do sensory systems build representations of complex features from simpler ones? Visual motion representation in cortex is a well-studied example: the direction and speed of moving objects, regardless of shape or texture, is computed from the local motion of oriented edges. Here we quantify tuning properties based on single-unit recordings in primate area MT, then fit a novel, generalized model of motion computation. The model reveals two core properties of MT neuronsspeed tuning and invariance to local edge orientation -result from a single organizing principle: each MT neuron combines afferents that represent edge motions consistent with a common velocity, much as V1 simple cells combine thalamic inputs consistent with a common orientation.

show abstract

Section: Conclusion Drawn From the Separable Modelmentioning

confidence: 99%

Compound stimuli reveal the structure of visual motion selectivity in macaque MT neurons

Zaharia

Goris

Movshon

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…We refer to this approach as neural system identification [Stanley, 2005, Wu et al, 2006. Neural system identification has been used to reveal mechanisms of neural information processing in biological systems [Joukes et al, 2014, Klindt et al, 2017, St-Yves and Naselaris, 2018, Antolík et al, 2016, McIntosh et al, 2016, Batty et al, 2017. However, so far these ideas have mostly been applied within individual brain regions.…”

Section: Introductionmentioning

confidence: 99%

Neural System Identification with Cortical Information Flow

Seeliger¹,

Ambrogioni²,

Güçlütürk³

et al. 2019

Preprint

View full text Add to dashboard Cite

These authors contributed equally to this work. AbstractNeural information flow (NIF) is a new framework for system identification in neuroscience. It integrates population receptive field estimation, neural encoding, connectivity analysis and hemodynamic response estimation in a single differentiable model that can be trained end-to-end via stochastic gradient descent. NIF models represent neural information processing systems as a network of coupled tensors, each encoding the representation of the sensory input contained in a brain region. The elements of these tensors can be interpreted as cortical columns whose activity encodes the presence of a specific feature in a spatiotemporal location. Each tensor is coupled to the measured data specific to a brain region via lowrank observation models that can be decomposed into the spatial, temporal and feature receptive fields of a localized neuronal population. Both these observation models and the convolutional weights defining the information processing within regions and effective connectivity between regions are learned end-to-end by predicting the neural signal during sensory stimulation. We trained a NIF model on the activity of early visual areas using a large-scale fMRI dataset. We show that we can recover plausible visual representations and population receptive fields that are consistent with empirical findings.

show abstract

“…This stimulation protocol is applicable to arbitrary stimulus, and can simply be extended to elementary temporal properties of V1 neural response by for example assuming space-time separability and expanding the RF model by a temporal filter reflecting the onset or offset dynamics. An interesting potential approach for setting the parameters of these filters would be to use reverse correlation approaches for their determination [68,9] directly from data. Furthermore, here we have restricted our consideration only to two stimulus feature selectivities present in V1 -the orientation and position.…”

mentioning

confidence: 99%

Cortical visual prosthesis: a detailed large-scale simulation study

Antolík

Sabatier

Galle

et al. 2019

Preprint

View full text Add to dashboard Cite

Recent advances in applying optogenetics in primates initiated the development of light based prosthetic implants for sensory restoration. Thanks to being the most well explored cortical area that is readily accessible at the surface of the brain, vision restoration via direct optogenetic activation of primary visual cortex is one of the most promising early targets for a optogenetics based prosthetic program. However, two fundamental elements of the cortical optogenetic prosthesis remain unclear. First, the exact mechanisms of neural dynamics under direct cortical stimulation, especially in the context of living, active and functionally specific intra-cortical neural circuitry, is poorly understood. Second, we lack protocols for transformation of arbitrary visual stimuli into light activation patterns that would induce perception of the said stimulus by the subject. In this study we address these issues using a large-scale spiking neural network modeling strategy of high biological fidelity. We examine the relationship between specific spatial configuration of light delivered to cortex and the resulting spatio-temporal pattern of activity evoked in the simulated cortical circuitry. Using such virtual experiments, we design a protocol for translation of a specific set of stimuli to activation pattern of a matrix of light emitting elements and provide a detailed assessment of the resulting cortical activations with respect to the natural vision control condition. In this study we restrict our focus to the grating stimulus class, which are an ideal starting point for exploration due to their thoroughly characterized representation in V1 and well-defined information content. However, we also provide an outline of a straight-forward road-map for transforming this grating centric stimulation protocol towards general strategy capable of transforming arbitrary spatio-temporal visual stimulus to a spatio-temporal pattern of light, thus enabling vision restoration via optogenetic V1 activation.

show abstract

Model Constrained by Visual Hierarchy Improves Prediction of Neural Responses to Natural Scenes

Cited by 55 publications

References 43 publications

Compound stimuli reveal the structure of visual motion selectivity in macaque MT neurons

Compound stimuli reveal the structure of visual motion selectivity in macaque MT neurons

Neural System Identification with Cortical Information Flow

Cortical visual prosthesis: a detailed large-scale simulation study

Contact Info

Product

Resources

About