Estimating sparse spectro-temporal receptive fields with natural stimuli

Stephen David, Nima Mesgarani, Shihab A. Shamma

Research output: Contribution to journalArticle

62 Citations (Scopus)

Abstract

Several algorithms have been proposed to characterize the spectro-temporal tuning properties of auditory neurons during the presentation of natural stimuli. Algorithms designed to work at realistic signal-to-noise levels must make some prior assumptions about tuning in order to produce accurate fits, and these priors can introduce bias into estimates of tuning. We compare a new, computationally efficient algorithm for estimating tuning properties, boosting, to a more commonly used algorithm, normalized reverse correlation. These algorithms employ the same functional model and cost function, differing only in their priors. We use both algorithms to estimate spectro-temporal tuning properties of neurons in primary auditory cortex during the presentation of continuous human speech. Models estimated using either algorithm, have similar predictive power, although fits by boosting are slightly more accurate. More strikingly, neurons characterized with boosting appear tuned to narrower spectral bandwidths and higher temporal modulation rates than when characterized with normalized reverse correlation. These differences have little impact on responses to speech, which is spectrally broadband and modulated at low rates. However, we find that models estimated by boosting also predict responses to non-speech stimuli more accurately. These findings highlight the crucial role of priors in characterizing neuronal response properties with natural stimuli.

Original languageEnglish (US)
Pages (from-to)191-212
Number of pages22
JournalNetwork: Computation in Neural Systems
Volume18
Issue number3
DOIs
StatePublished - 2007
Externally publishedYes

Fingerprint

Neurons
Auditory Cortex
Noise
Costs and Cost Analysis

Keywords

  • Auditory cortex
  • Boosting
  • Reverse correlation
  • Speech

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Estimating sparse spectro-temporal receptive fields with natural stimuli. / David, Stephen; Mesgarani, Nima; Shamma, Shihab A.

In: Network: Computation in Neural Systems, Vol. 18, No. 3, 2007, p. 191-212.

Research output: Contribution to journalArticle

David, Stephen ; Mesgarani, Nima ; Shamma, Shihab A. / Estimating sparse spectro-temporal receptive fields with natural stimuli. In: Network: Computation in Neural Systems. 2007 ; Vol. 18, No. 3. pp. 191-212.
@article{0c07ade598ed4993a6a4f2843adc9ac7,
title = "Estimating sparse spectro-temporal receptive fields with natural stimuli",
abstract = "Several algorithms have been proposed to characterize the spectro-temporal tuning properties of auditory neurons during the presentation of natural stimuli. Algorithms designed to work at realistic signal-to-noise levels must make some prior assumptions about tuning in order to produce accurate fits, and these priors can introduce bias into estimates of tuning. We compare a new, computationally efficient algorithm for estimating tuning properties, boosting, to a more commonly used algorithm, normalized reverse correlation. These algorithms employ the same functional model and cost function, differing only in their priors. We use both algorithms to estimate spectro-temporal tuning properties of neurons in primary auditory cortex during the presentation of continuous human speech. Models estimated using either algorithm, have similar predictive power, although fits by boosting are slightly more accurate. More strikingly, neurons characterized with boosting appear tuned to narrower spectral bandwidths and higher temporal modulation rates than when characterized with normalized reverse correlation. These differences have little impact on responses to speech, which is spectrally broadband and modulated at low rates. However, we find that models estimated by boosting also predict responses to non-speech stimuli more accurately. These findings highlight the crucial role of priors in characterizing neuronal response properties with natural stimuli.",
keywords = "Auditory cortex, Boosting, Reverse correlation, Speech",
author = "Stephen David and Nima Mesgarani and Shamma, {Shihab A.}",
year = "2007",
doi = "10.1080/09548980701609235",
language = "English (US)",
volume = "18",
pages = "191--212",
journal = "Network: Computation in Neural Systems",
issn = "0954-898X",
publisher = "Informa Healthcare",
number = "3",

}

TY - JOUR

T1 - Estimating sparse spectro-temporal receptive fields with natural stimuli

AU - David, Stephen

AU - Mesgarani, Nima

AU - Shamma, Shihab A.

PY - 2007

Y1 - 2007

N2 - Several algorithms have been proposed to characterize the spectro-temporal tuning properties of auditory neurons during the presentation of natural stimuli. Algorithms designed to work at realistic signal-to-noise levels must make some prior assumptions about tuning in order to produce accurate fits, and these priors can introduce bias into estimates of tuning. We compare a new, computationally efficient algorithm for estimating tuning properties, boosting, to a more commonly used algorithm, normalized reverse correlation. These algorithms employ the same functional model and cost function, differing only in their priors. We use both algorithms to estimate spectro-temporal tuning properties of neurons in primary auditory cortex during the presentation of continuous human speech. Models estimated using either algorithm, have similar predictive power, although fits by boosting are slightly more accurate. More strikingly, neurons characterized with boosting appear tuned to narrower spectral bandwidths and higher temporal modulation rates than when characterized with normalized reverse correlation. These differences have little impact on responses to speech, which is spectrally broadband and modulated at low rates. However, we find that models estimated by boosting also predict responses to non-speech stimuli more accurately. These findings highlight the crucial role of priors in characterizing neuronal response properties with natural stimuli.

AB - Several algorithms have been proposed to characterize the spectro-temporal tuning properties of auditory neurons during the presentation of natural stimuli. Algorithms designed to work at realistic signal-to-noise levels must make some prior assumptions about tuning in order to produce accurate fits, and these priors can introduce bias into estimates of tuning. We compare a new, computationally efficient algorithm for estimating tuning properties, boosting, to a more commonly used algorithm, normalized reverse correlation. These algorithms employ the same functional model and cost function, differing only in their priors. We use both algorithms to estimate spectro-temporal tuning properties of neurons in primary auditory cortex during the presentation of continuous human speech. Models estimated using either algorithm, have similar predictive power, although fits by boosting are slightly more accurate. More strikingly, neurons characterized with boosting appear tuned to narrower spectral bandwidths and higher temporal modulation rates than when characterized with normalized reverse correlation. These differences have little impact on responses to speech, which is spectrally broadband and modulated at low rates. However, we find that models estimated by boosting also predict responses to non-speech stimuli more accurately. These findings highlight the crucial role of priors in characterizing neuronal response properties with natural stimuli.

KW - Auditory cortex

KW - Boosting

KW - Reverse correlation

KW - Speech

UR - http://www.scopus.com/inward/record.url?scp=35148869003&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=35148869003&partnerID=8YFLogxK

U2 - 10.1080/09548980701609235

DO - 10.1080/09548980701609235

M3 - Article

C2 - 17852750

AN - SCOPUS:35148869003

VL - 18

SP - 191

EP - 212

JO - Network: Computation in Neural Systems

JF - Network: Computation in Neural Systems

SN - 0954-898X

IS - 3

ER -