Dimensionality reduction in neural models: An information-theoretic generalization of spike-triggered average and covariance analysis

Jonathan W. Pillow, Eero Simoncelli

Research output: Contribution to journalArticle

Abstract

We describe an information-theoretic framework for fitting neural spike responses with a Linear-Nonlinear-Poisson cascade model. This framework unifies the spike-triggered average (STA) and spike-triggered covariance (STC) approaches to neural characterization and recovers a set of linear filters that maximize mean and variance-dependent information between stimuli and spike responses. The resulting approach has several useful properties, namely, (1) it recovers a set of linear filters sorted according to their informativeness about the neural response; (2) it is both computationally efficient and robust, allowing recovery of multiple linear filters from a data set of relatively modest size; (3) it provides an explicit "default" model of the nonlinear stage mapping the filter responses to spike rate, in the form of a ratio of Gaussians; (4) it is equivalent to maximum likelihood estimation of this default model but also converges to the correct filter estimates whenever the conditions for the consistency of STA or STC analysis are met; and (5) it can be augmented with additional constraints on the filters, such as space-time separability. We demonstrate the effectiveness of the method by applying it to simulated responses of a Hodgkin-Huxley neuron and the recorded extracellular responses of macaque retinal ganglion cells and V1 cells.

Original languageEnglish (US)
Article number9
Pages (from-to)414-428
Number of pages15
JournalJournal of Vision
Volume6
Issue number4
DOIs
StatePublished - Apr 28 2006

Fingerprint

Retinal Ganglion Cells
Macaca
Neurons
Datasets

Keywords

  • Information theory
  • Neural coding
  • Neural modeling
  • Receptive field
  • Reverse correlation
  • White noise analysis

ASJC Scopus subject areas

  • Ophthalmology

Cite this

Dimensionality reduction in neural models : An information-theoretic generalization of spike-triggered average and covariance analysis. / Pillow, Jonathan W.; Simoncelli, Eero.

In: Journal of Vision, Vol. 6, No. 4, 9, 28.04.2006, p. 414-428.

Research output: Contribution to journalArticle

@article{1f7ebd955b3b40578ce2a50155a2c693,
title = "Dimensionality reduction in neural models: An information-theoretic generalization of spike-triggered average and covariance analysis",
abstract = "We describe an information-theoretic framework for fitting neural spike responses with a Linear-Nonlinear-Poisson cascade model. This framework unifies the spike-triggered average (STA) and spike-triggered covariance (STC) approaches to neural characterization and recovers a set of linear filters that maximize mean and variance-dependent information between stimuli and spike responses. The resulting approach has several useful properties, namely, (1) it recovers a set of linear filters sorted according to their informativeness about the neural response; (2) it is both computationally efficient and robust, allowing recovery of multiple linear filters from a data set of relatively modest size; (3) it provides an explicit {"}default{"} model of the nonlinear stage mapping the filter responses to spike rate, in the form of a ratio of Gaussians; (4) it is equivalent to maximum likelihood estimation of this default model but also converges to the correct filter estimates whenever the conditions for the consistency of STA or STC analysis are met; and (5) it can be augmented with additional constraints on the filters, such as space-time separability. We demonstrate the effectiveness of the method by applying it to simulated responses of a Hodgkin-Huxley neuron and the recorded extracellular responses of macaque retinal ganglion cells and V1 cells.",
keywords = "Information theory, Neural coding, Neural modeling, Receptive field, Reverse correlation, White noise analysis",
author = "Pillow, {Jonathan W.} and Eero Simoncelli",
year = "2006",
month = "4",
day = "28",
doi = "10.1167/6.4.9",
language = "English (US)",
volume = "6",
pages = "414--428",
journal = "Journal of Vision",
issn = "1534-7362",
publisher = "Association for Research in Vision and Ophthalmology Inc.",
number = "4",

}

TY - JOUR

T1 - Dimensionality reduction in neural models

T2 - An information-theoretic generalization of spike-triggered average and covariance analysis

AU - Pillow, Jonathan W.

AU - Simoncelli, Eero

PY - 2006/4/28

Y1 - 2006/4/28

N2 - We describe an information-theoretic framework for fitting neural spike responses with a Linear-Nonlinear-Poisson cascade model. This framework unifies the spike-triggered average (STA) and spike-triggered covariance (STC) approaches to neural characterization and recovers a set of linear filters that maximize mean and variance-dependent information between stimuli and spike responses. The resulting approach has several useful properties, namely, (1) it recovers a set of linear filters sorted according to their informativeness about the neural response; (2) it is both computationally efficient and robust, allowing recovery of multiple linear filters from a data set of relatively modest size; (3) it provides an explicit "default" model of the nonlinear stage mapping the filter responses to spike rate, in the form of a ratio of Gaussians; (4) it is equivalent to maximum likelihood estimation of this default model but also converges to the correct filter estimates whenever the conditions for the consistency of STA or STC analysis are met; and (5) it can be augmented with additional constraints on the filters, such as space-time separability. We demonstrate the effectiveness of the method by applying it to simulated responses of a Hodgkin-Huxley neuron and the recorded extracellular responses of macaque retinal ganglion cells and V1 cells.

AB - We describe an information-theoretic framework for fitting neural spike responses with a Linear-Nonlinear-Poisson cascade model. This framework unifies the spike-triggered average (STA) and spike-triggered covariance (STC) approaches to neural characterization and recovers a set of linear filters that maximize mean and variance-dependent information between stimuli and spike responses. The resulting approach has several useful properties, namely, (1) it recovers a set of linear filters sorted according to their informativeness about the neural response; (2) it is both computationally efficient and robust, allowing recovery of multiple linear filters from a data set of relatively modest size; (3) it provides an explicit "default" model of the nonlinear stage mapping the filter responses to spike rate, in the form of a ratio of Gaussians; (4) it is equivalent to maximum likelihood estimation of this default model but also converges to the correct filter estimates whenever the conditions for the consistency of STA or STC analysis are met; and (5) it can be augmented with additional constraints on the filters, such as space-time separability. We demonstrate the effectiveness of the method by applying it to simulated responses of a Hodgkin-Huxley neuron and the recorded extracellular responses of macaque retinal ganglion cells and V1 cells.

KW - Information theory

KW - Neural coding

KW - Neural modeling

KW - Receptive field

KW - Reverse correlation

KW - White noise analysis

UR - http://www.scopus.com/inward/record.url?scp=33646673326&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646673326&partnerID=8YFLogxK

U2 - 10.1167/6.4.9

DO - 10.1167/6.4.9

M3 - Article

C2 - 16889478

AN - SCOPUS:33646673326

VL - 6

SP - 414

EP - 428

JO - Journal of Vision

JF - Journal of Vision

SN - 1534-7362

IS - 4

M1 - 9

ER -