Maximum likelihood estimation of a stochastic integrate-and-fire neural encoding model

Liam Paninski, Jonathan W. Pillow, Eero Simoncelli

Research output: Contribution to journalArticle

Abstract

We examine a cascade encoding model for neural response in which a linear filtering stage is followed by a noisy, leaky, integrate-and-fire spike generation mechanism. This model provides a biophysically more realistic alternative to models based on Poisson (memoryless) spike generation, and can effectively reproduce a variety of spiking behaviors seen in vivo. We describe the maximum likelihood estimator for the model parameters, given only extracellular spike train responses (not intracellular voltage data). Specifically, we prove that the log-likelihood function is concave and thus has an essentially unique global maximum that can be found using gradient ascent techniques. We develop an efficient algorithm for computing the maximum likelihood solution, demonstrate the effectiveness of the resulting estimator with numerical simulations, and discuss a method of testing the model's validity using time-rescaling and density evolution techniques.

Original languageEnglish (US)
Pages (from-to)2533-2561
Number of pages29
JournalNeural Computation
Volume16
Issue number12
DOIs
StatePublished - Dec 2004

Fingerprint

Likelihood Functions
Maximum likelihood estimation
Fires
Maximum likelihood
Computer simulation
Testing
Electric potential

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Artificial Intelligence
  • Neuroscience(all)

Cite this

Maximum likelihood estimation of a stochastic integrate-and-fire neural encoding model. / Paninski, Liam; Pillow, Jonathan W.; Simoncelli, Eero.

In: Neural Computation, Vol. 16, No. 12, 12.2004, p. 2533-2561.

Research output: Contribution to journalArticle

Paninski, Liam ; Pillow, Jonathan W. ; Simoncelli, Eero. / Maximum likelihood estimation of a stochastic integrate-and-fire neural encoding model. In: Neural Computation. 2004 ; Vol. 16, No. 12. pp. 2533-2561.
@article{8d2f2e4c6aeb41bf83625bbda594200c,
title = "Maximum likelihood estimation of a stochastic integrate-and-fire neural encoding model",
abstract = "We examine a cascade encoding model for neural response in which a linear filtering stage is followed by a noisy, leaky, integrate-and-fire spike generation mechanism. This model provides a biophysically more realistic alternative to models based on Poisson (memoryless) spike generation, and can effectively reproduce a variety of spiking behaviors seen in vivo. We describe the maximum likelihood estimator for the model parameters, given only extracellular spike train responses (not intracellular voltage data). Specifically, we prove that the log-likelihood function is concave and thus has an essentially unique global maximum that can be found using gradient ascent techniques. We develop an efficient algorithm for computing the maximum likelihood solution, demonstrate the effectiveness of the resulting estimator with numerical simulations, and discuss a method of testing the model's validity using time-rescaling and density evolution techniques.",
author = "Liam Paninski and Pillow, {Jonathan W.} and Eero Simoncelli",
year = "2004",
month = "12",
doi = "10.1162/0899766042321797",
language = "English (US)",
volume = "16",
pages = "2533--2561",
journal = "Neural Computation",
issn = "0899-7667",
publisher = "MIT Press Journals",
number = "12",

}

TY - JOUR

T1 - Maximum likelihood estimation of a stochastic integrate-and-fire neural encoding model

AU - Paninski, Liam

AU - Pillow, Jonathan W.

AU - Simoncelli, Eero

PY - 2004/12

Y1 - 2004/12

N2 - We examine a cascade encoding model for neural response in which a linear filtering stage is followed by a noisy, leaky, integrate-and-fire spike generation mechanism. This model provides a biophysically more realistic alternative to models based on Poisson (memoryless) spike generation, and can effectively reproduce a variety of spiking behaviors seen in vivo. We describe the maximum likelihood estimator for the model parameters, given only extracellular spike train responses (not intracellular voltage data). Specifically, we prove that the log-likelihood function is concave and thus has an essentially unique global maximum that can be found using gradient ascent techniques. We develop an efficient algorithm for computing the maximum likelihood solution, demonstrate the effectiveness of the resulting estimator with numerical simulations, and discuss a method of testing the model's validity using time-rescaling and density evolution techniques.

AB - We examine a cascade encoding model for neural response in which a linear filtering stage is followed by a noisy, leaky, integrate-and-fire spike generation mechanism. This model provides a biophysically more realistic alternative to models based on Poisson (memoryless) spike generation, and can effectively reproduce a variety of spiking behaviors seen in vivo. We describe the maximum likelihood estimator for the model parameters, given only extracellular spike train responses (not intracellular voltage data). Specifically, we prove that the log-likelihood function is concave and thus has an essentially unique global maximum that can be found using gradient ascent techniques. We develop an efficient algorithm for computing the maximum likelihood solution, demonstrate the effectiveness of the resulting estimator with numerical simulations, and discuss a method of testing the model's validity using time-rescaling and density evolution techniques.

UR - http://www.scopus.com/inward/record.url?scp=9744239998&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=9744239998&partnerID=8YFLogxK

U2 - 10.1162/0899766042321797

DO - 10.1162/0899766042321797

M3 - Article

VL - 16

SP - 2533

EP - 2561

JO - Neural Computation

JF - Neural Computation

SN - 0899-7667

IS - 12

ER -