Spiking networks for Bayesian inference and choice

Wei Ji Ma, Jeffrey M. Beck, Alexandre Pouget

Research output: Contribution to journalArticle

Abstract

Systems neuroscience traditionally conceptualizes a population of spiking neurons as merely encoding the value of a stimulus. Yet, psychophysics has revealed that people take into account stimulus uncertainty when performing sensory or motor computations and do so in a nearly Bayes-optimal way. This suggests that neural populations do not encode just a single value but an entire probability distribution over the stimulus. Several such probabilistic codes have been proposed, including one that utilizes the structure of neural variability to enable simple neural implementations of probabilistic computations such as optimal cue integration. This approach provides a quantitative link between Bayes-optimal behaviors and specific neural operations. It allows for novel ways to evaluate probabilistic codes and for predictions for physiological population recordings.

Original languageEnglish (US)
Pages (from-to)217-222
Number of pages6
JournalCurrent Opinion in Neurobiology
Volume18
Issue number2
DOIs
StatePublished - Apr 2008

Fingerprint

Psychophysics
Population
Neurosciences
Uncertainty
Cues
Neurons

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Spiking networks for Bayesian inference and choice. / Ma, Wei Ji; Beck, Jeffrey M.; Pouget, Alexandre.

In: Current Opinion in Neurobiology, Vol. 18, No. 2, 04.2008, p. 217-222.

Research output: Contribution to journalArticle

Ma, Wei Ji ; Beck, Jeffrey M. ; Pouget, Alexandre. / Spiking networks for Bayesian inference and choice. In: Current Opinion in Neurobiology. 2008 ; Vol. 18, No. 2. pp. 217-222.
@article{070fec58ab174a89ac394b6869e14ae9,
title = "Spiking networks for Bayesian inference and choice",
abstract = "Systems neuroscience traditionally conceptualizes a population of spiking neurons as merely encoding the value of a stimulus. Yet, psychophysics has revealed that people take into account stimulus uncertainty when performing sensory or motor computations and do so in a nearly Bayes-optimal way. This suggests that neural populations do not encode just a single value but an entire probability distribution over the stimulus. Several such probabilistic codes have been proposed, including one that utilizes the structure of neural variability to enable simple neural implementations of probabilistic computations such as optimal cue integration. This approach provides a quantitative link between Bayes-optimal behaviors and specific neural operations. It allows for novel ways to evaluate probabilistic codes and for predictions for physiological population recordings.",
author = "Ma, {Wei Ji} and Beck, {Jeffrey M.} and Alexandre Pouget",
year = "2008",
month = "4",
doi = "10.1016/j.conb.2008.07.004",
language = "English (US)",
volume = "18",
pages = "217--222",
journal = "Current Opinion in Neurobiology",
issn = "0959-4388",
publisher = "Elsevier Limited",
number = "2",

}

TY - JOUR

T1 - Spiking networks for Bayesian inference and choice

AU - Ma, Wei Ji

AU - Beck, Jeffrey M.

AU - Pouget, Alexandre

PY - 2008/4

Y1 - 2008/4

N2 - Systems neuroscience traditionally conceptualizes a population of spiking neurons as merely encoding the value of a stimulus. Yet, psychophysics has revealed that people take into account stimulus uncertainty when performing sensory or motor computations and do so in a nearly Bayes-optimal way. This suggests that neural populations do not encode just a single value but an entire probability distribution over the stimulus. Several such probabilistic codes have been proposed, including one that utilizes the structure of neural variability to enable simple neural implementations of probabilistic computations such as optimal cue integration. This approach provides a quantitative link between Bayes-optimal behaviors and specific neural operations. It allows for novel ways to evaluate probabilistic codes and for predictions for physiological population recordings.

AB - Systems neuroscience traditionally conceptualizes a population of spiking neurons as merely encoding the value of a stimulus. Yet, psychophysics has revealed that people take into account stimulus uncertainty when performing sensory or motor computations and do so in a nearly Bayes-optimal way. This suggests that neural populations do not encode just a single value but an entire probability distribution over the stimulus. Several such probabilistic codes have been proposed, including one that utilizes the structure of neural variability to enable simple neural implementations of probabilistic computations such as optimal cue integration. This approach provides a quantitative link between Bayes-optimal behaviors and specific neural operations. It allows for novel ways to evaluate probabilistic codes and for predictions for physiological population recordings.

UR - http://www.scopus.com/inward/record.url?scp=52049084777&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=52049084777&partnerID=8YFLogxK

U2 - 10.1016/j.conb.2008.07.004

DO - 10.1016/j.conb.2008.07.004

M3 - Article

VL - 18

SP - 217

EP - 222

JO - Current Opinion in Neurobiology

JF - Current Opinion in Neurobiology

SN - 0959-4388

IS - 2

ER -