Predicting neuronal responses during natural vision

Stephen David, Jack L. Gallant

Research output: Contribution to journalArticle

90 Citations (Scopus)

Abstract

A model that fully describes the response properties of visual neurons must be able to predict their activity during natural vision. While many models have been proposed for the visual system, few have ever been tested against this criterion. To address this issue, we have developed a general framework for fitting and validating nonlinear models of visual neurons using natural visual stimuli. Our approach derives from linear spatiotemporal receptive field (STRF) analysis, which has frequently been used to study the visual system. However, prior to the linear filtering stage typical of STRFs, a linearizing transformation is applied to the stimulus to account for nonlinear response properties. We used this approach to compare two models for neurons in primary visual cortex: a nonlinear Fourier power model, which accounts for spatial phase invariant tuning, and a traditional linear model. We characterized prediction accuracy in terms of the total explainable variance, given intrinsic experimental noise. On average, Fourier power STRFs predicted 40% of explainable variance while linear STRFs were able to predict only 21% of explainable variance. The performance of the Fourier power model provides a benchmark for evaluating more sophisticated models in the future.

Original languageEnglish (US)
Pages (from-to)239-260
Number of pages22
JournalNetwork: Computation in Neural Systems
Volume16
Issue number2-3
DOIs
StatePublished - Jun 2005
Externally publishedYes

Fingerprint

Neurons
Benchmarking
Nonlinear Dynamics
Visual Cortex
Noise
Linear Models
serum thyroid hormone reducing factor
Power (Psychology)

Keywords

  • Nonlinear model
  • Prediction
  • Receptive field
  • Visual cortex

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Predicting neuronal responses during natural vision. / David, Stephen; Gallant, Jack L.

In: Network: Computation in Neural Systems, Vol. 16, No. 2-3, 06.2005, p. 239-260.

Research output: Contribution to journalArticle

@article{b14bd3f3218c4cf59e935a7666d27655,
title = "Predicting neuronal responses during natural vision",
abstract = "A model that fully describes the response properties of visual neurons must be able to predict their activity during natural vision. While many models have been proposed for the visual system, few have ever been tested against this criterion. To address this issue, we have developed a general framework for fitting and validating nonlinear models of visual neurons using natural visual stimuli. Our approach derives from linear spatiotemporal receptive field (STRF) analysis, which has frequently been used to study the visual system. However, prior to the linear filtering stage typical of STRFs, a linearizing transformation is applied to the stimulus to account for nonlinear response properties. We used this approach to compare two models for neurons in primary visual cortex: a nonlinear Fourier power model, which accounts for spatial phase invariant tuning, and a traditional linear model. We characterized prediction accuracy in terms of the total explainable variance, given intrinsic experimental noise. On average, Fourier power STRFs predicted 40{\%} of explainable variance while linear STRFs were able to predict only 21{\%} of explainable variance. The performance of the Fourier power model provides a benchmark for evaluating more sophisticated models in the future.",
keywords = "Nonlinear model, Prediction, Receptive field, Visual cortex",
author = "Stephen David and Gallant, {Jack L.}",
year = "2005",
month = "6",
doi = "10.1080/09548980500464030",
language = "English (US)",
volume = "16",
pages = "239--260",
journal = "Network: Computation in Neural Systems",
issn = "0954-898X",
publisher = "Informa Healthcare",
number = "2-3",

}

TY - JOUR

T1 - Predicting neuronal responses during natural vision

AU - David, Stephen

AU - Gallant, Jack L.

PY - 2005/6

Y1 - 2005/6

N2 - A model that fully describes the response properties of visual neurons must be able to predict their activity during natural vision. While many models have been proposed for the visual system, few have ever been tested against this criterion. To address this issue, we have developed a general framework for fitting and validating nonlinear models of visual neurons using natural visual stimuli. Our approach derives from linear spatiotemporal receptive field (STRF) analysis, which has frequently been used to study the visual system. However, prior to the linear filtering stage typical of STRFs, a linearizing transformation is applied to the stimulus to account for nonlinear response properties. We used this approach to compare two models for neurons in primary visual cortex: a nonlinear Fourier power model, which accounts for spatial phase invariant tuning, and a traditional linear model. We characterized prediction accuracy in terms of the total explainable variance, given intrinsic experimental noise. On average, Fourier power STRFs predicted 40% of explainable variance while linear STRFs were able to predict only 21% of explainable variance. The performance of the Fourier power model provides a benchmark for evaluating more sophisticated models in the future.

AB - A model that fully describes the response properties of visual neurons must be able to predict their activity during natural vision. While many models have been proposed for the visual system, few have ever been tested against this criterion. To address this issue, we have developed a general framework for fitting and validating nonlinear models of visual neurons using natural visual stimuli. Our approach derives from linear spatiotemporal receptive field (STRF) analysis, which has frequently been used to study the visual system. However, prior to the linear filtering stage typical of STRFs, a linearizing transformation is applied to the stimulus to account for nonlinear response properties. We used this approach to compare two models for neurons in primary visual cortex: a nonlinear Fourier power model, which accounts for spatial phase invariant tuning, and a traditional linear model. We characterized prediction accuracy in terms of the total explainable variance, given intrinsic experimental noise. On average, Fourier power STRFs predicted 40% of explainable variance while linear STRFs were able to predict only 21% of explainable variance. The performance of the Fourier power model provides a benchmark for evaluating more sophisticated models in the future.

KW - Nonlinear model

KW - Prediction

KW - Receptive field

KW - Visual cortex

UR - http://www.scopus.com/inward/record.url?scp=30444433256&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=30444433256&partnerID=8YFLogxK

U2 - 10.1080/09548980500464030

DO - 10.1080/09548980500464030

M3 - Article

C2 - 16411498

AN - SCOPUS:30444433256

VL - 16

SP - 239

EP - 260

JO - Network: Computation in Neural Systems

JF - Network: Computation in Neural Systems

SN - 0954-898X

IS - 2-3

ER -