Nonlinear V1 responses to natural scenes revealed by neural network analysis

Ryan Prenger, Michael C K Wu, Stephen David, Jack L. Gallant

Research output: Contribution to journalArticle

46 Citations (Scopus)

Abstract

A key goal in the study of visual processing is to obtain a comprehensive description of the relationship between visual stimuli and neuronal responses. One way to guide the search for models is to use a general nonparametric regression algorithm, such as a neural network. We have developed a multilayer feed-forward network algorithm that can be used to characterize nonlinear stimulus-response mapping functions of neurons in primary visual cortex (area V1) using natural image stimuli. The network is capable of extracting several known V1 response properties such as: orientation and spatial frequency tuning, the spatial phase invariance of complex cells, and direction selectivity. We present details of a method for training networks and visualizing their properties. We also compare how well conventional explicit models and those developed using neural networks can predict novel responses to natural scenes.

Original languageEnglish (US)
Pages (from-to)663-679
Number of pages17
JournalNeural Networks
Volume17
Issue number5-6
DOIs
StatePublished - Jun 2004
Externally publishedYes

Fingerprint

Electric network analysis
Neural networks
Visual Cortex
Invariance
Neurons
Multilayers
Tuning
Processing
Direction compound

Keywords

  • Multi-layer perceptron
  • Natural vision
  • Nonparametric model
  • Prediction
  • Receptive field
  • Reverse correlation
  • Striate cortex

ASJC Scopus subject areas

  • Artificial Intelligence
  • Neuroscience(all)

Cite this

Nonlinear V1 responses to natural scenes revealed by neural network analysis. / Prenger, Ryan; Wu, Michael C K; David, Stephen; Gallant, Jack L.

In: Neural Networks, Vol. 17, No. 5-6, 06.2004, p. 663-679.

Research output: Contribution to journalArticle

Prenger, Ryan ; Wu, Michael C K ; David, Stephen ; Gallant, Jack L. / Nonlinear V1 responses to natural scenes revealed by neural network analysis. In: Neural Networks. 2004 ; Vol. 17, No. 5-6. pp. 663-679.
@article{acd8f342da714aeba97fd8f69a1b45cd,
title = "Nonlinear V1 responses to natural scenes revealed by neural network analysis",
abstract = "A key goal in the study of visual processing is to obtain a comprehensive description of the relationship between visual stimuli and neuronal responses. One way to guide the search for models is to use a general nonparametric regression algorithm, such as a neural network. We have developed a multilayer feed-forward network algorithm that can be used to characterize nonlinear stimulus-response mapping functions of neurons in primary visual cortex (area V1) using natural image stimuli. The network is capable of extracting several known V1 response properties such as: orientation and spatial frequency tuning, the spatial phase invariance of complex cells, and direction selectivity. We present details of a method for training networks and visualizing their properties. We also compare how well conventional explicit models and those developed using neural networks can predict novel responses to natural scenes.",
keywords = "Multi-layer perceptron, Natural vision, Nonparametric model, Prediction, Receptive field, Reverse correlation, Striate cortex",
author = "Ryan Prenger and Wu, {Michael C K} and Stephen David and Gallant, {Jack L.}",
year = "2004",
month = "6",
doi = "10.1016/j.neunet.2004.03.008",
language = "English (US)",
volume = "17",
pages = "663--679",
journal = "Neural Networks",
issn = "0893-6080",
publisher = "Elsevier Limited",
number = "5-6",

}

TY - JOUR

T1 - Nonlinear V1 responses to natural scenes revealed by neural network analysis

AU - Prenger, Ryan

AU - Wu, Michael C K

AU - David, Stephen

AU - Gallant, Jack L.

PY - 2004/6

Y1 - 2004/6

N2 - A key goal in the study of visual processing is to obtain a comprehensive description of the relationship between visual stimuli and neuronal responses. One way to guide the search for models is to use a general nonparametric regression algorithm, such as a neural network. We have developed a multilayer feed-forward network algorithm that can be used to characterize nonlinear stimulus-response mapping functions of neurons in primary visual cortex (area V1) using natural image stimuli. The network is capable of extracting several known V1 response properties such as: orientation and spatial frequency tuning, the spatial phase invariance of complex cells, and direction selectivity. We present details of a method for training networks and visualizing their properties. We also compare how well conventional explicit models and those developed using neural networks can predict novel responses to natural scenes.

AB - A key goal in the study of visual processing is to obtain a comprehensive description of the relationship between visual stimuli and neuronal responses. One way to guide the search for models is to use a general nonparametric regression algorithm, such as a neural network. We have developed a multilayer feed-forward network algorithm that can be used to characterize nonlinear stimulus-response mapping functions of neurons in primary visual cortex (area V1) using natural image stimuli. The network is capable of extracting several known V1 response properties such as: orientation and spatial frequency tuning, the spatial phase invariance of complex cells, and direction selectivity. We present details of a method for training networks and visualizing their properties. We also compare how well conventional explicit models and those developed using neural networks can predict novel responses to natural scenes.

KW - Multi-layer perceptron

KW - Natural vision

KW - Nonparametric model

KW - Prediction

KW - Receptive field

KW - Reverse correlation

KW - Striate cortex

UR - http://www.scopus.com/inward/record.url?scp=4344612285&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=4344612285&partnerID=8YFLogxK

U2 - 10.1016/j.neunet.2004.03.008

DO - 10.1016/j.neunet.2004.03.008

M3 - Article

C2 - 15288891

AN - SCOPUS:4344612285

VL - 17

SP - 663

EP - 679

JO - Neural Networks

JF - Neural Networks

SN - 0893-6080

IS - 5-6

ER -