A perturbative approach to model retinal processing of natural scenes

This project aims to develop realistic deep network models to understand retinal processing of natural scenes by mapping model components to retinal cell types and probing selectivity to stimuli perturbations.

Subsidie
€ 1.998.280
2022

Projectdetails

Introduction

A major goal of sensory neuroscience is to understand how sensory neurons process natural scenes. Models built from the responses of sensory neurons to simple stimuli do not generalize to predict how complex natural scenes are processed. Even as early as in the retina, this issue is not solved.

Deep Network Models

Deep network models have been proposed to predict the responses of visual neurons to natural stimuli. However, they are still far from being a realistic model of the visual system.

Limitations of Current Models

  1. The sensitivity to perturbations of the stimulus can thus be very different for a deep network model and for our visual system.
  2. It is not clear how the model components can be related to actual mechanisms in the brain.

Project Purpose

Our purpose is to understand how the retina processes natural scenes. We will follow an interdisciplinary approach where we will build realistic deep network models of retinal processing and test them in experiments.

Model Development

We will develop deep network models that can predict ganglion cell responses to natural stimuli and map the components of these models to specific cell types in the retinal network.

Originality of the Project

Our project is original because it will use two novel methods that will be key to achieving our goal:

  1. A novel approach to characterize retinal function, where we will probe the selectivity of the retina to perturbations of natural stimuli.
  2. A novel tool based on 2-photon holographic stimulation to decompose the retinal circuit.

These methods are tailored to address the specific issues of deep networks.

Ganglion Cell Receptive Fields

Each ganglion cell has a receptive field center, the region of visual space whose stimulation evokes the strongest responses.

Project Structure

Our project is divided into three parts:

  1. We will first understand how natural images are integrated inside the receptive field center.
  2. We will then ask how stimulation outside the receptive field center affects ganglion cell processing of natural images.
  3. Finally, we will focus on motion processing during natural scene stimulation.

Financiële details & Tijdlijn

Financiële details

Subsidiebedrag€ 1.998.280
Totale projectbegroting€ 1.998.280

Tijdlijn

Startdatum1-10-2022
Einddatum30-9-2027
Subsidiejaar2022

Partners & Locaties

Projectpartners

  • INSTITUT NATIONAL DE LA SANTE ET DE LA RECHERCHE MEDICALEpenvoerder

Land(en)

France

Vergelijkbare projecten binnen European Research Council

ERC Starting...

Next generation mechanistic models of retinal interneurons

This project aims to develop hybrid mechanistic models of retinal amacrine cells, integrating machine learning and imaging data to uncover their roles in visual computations and link them to genetic types.

€ 1.499.860
ERC Starting...

Tracing Visual Computations from the Retina to Behavior

This project aims to investigate how the superior colliculus integrates retinal signals to drive behavior using imaging, optogenetics, and modeling, revealing mechanisms of visual information processing.

€ 1.871.465
ERC Starting...

Using deep neural networks to understand functional specialization in the human visual cortex

This project aims to uncover the origins of functional specialization in the brain's visual pathway by integrating computational modeling, naturalistic behavior sampling, and neuroimaging.

€ 1.494.750
ERC Starting...

Detailed Cortical Mechanisms of Top-Down Visual Processing

This project aims to explore the neural mechanisms of generative vision in macaque monkeys using advanced imaging and behavioral tasks, linking findings to artificial intelligence and human perception.

€ 1.500.000
ERC Starting...

It's about time: Towards a dynamic account of natural vision.

TIME aims to revolutionize vision research by integrating semantic understanding and active information sampling through advanced brain imaging and bio-inspired deep learning, enhancing insights into visual cognition.

€ 1.499.455