Synaptic computation underlying probabilistic inference

Alireza Soltani, Xiao-Jing Wang

Research output: Contribution to journalArticle

Abstract

We propose that synapses may be the workhorse of the neuronal computations that underlie probabilistic reasoning. We built a neural circuit model for probabilistic inference in which information provided by different sensory cues must be integrated and the predictive powers of individual cues about an outcome are deduced through experience. We found that bounded synapses naturally compute, through reward-dependent plasticity, the posterior probability that a choice alternative is correct given that a cue is presented. Furthermore, a decision circuit endowed with such synapses makes choices on the basis of the summed log posterior odds and performs near-optimal cue combination. The model was validated by reproducing salient observations of, and provides insights into, a monkey experiment using a categorization task. Our model thus suggests a biophysical instantiation of the Bayesian decision rule, while predicting important deviations from it similar to the 'base-rate neglect' observed in human studies when alternatives have unequal prior probabilities.

Original languageEnglish (US)
Pages (from-to)112-119
Number of pages8
JournalNature Neuroscience
Volume13
Issue number1
DOIs
StatePublished - Jan 17 2010

Fingerprint

Cues
Synapses
Statistical Models
Reward
Haplorhini

ASJC Scopus subject areas

  • Neuroscience(all)

Cite this

Synaptic computation underlying probabilistic inference. / Soltani, Alireza; Wang, Xiao-Jing.

In: Nature Neuroscience, Vol. 13, No. 1, 17.01.2010, p. 112-119.

Research output: Contribution to journalArticle

Soltani, Alireza ; Wang, Xiao-Jing. / Synaptic computation underlying probabilistic inference. In: Nature Neuroscience. 2010 ; Vol. 13, No. 1. pp. 112-119.
@article{2fef05e7278b458c8b1c79870b1dae5f,
title = "Synaptic computation underlying probabilistic inference",
abstract = "We propose that synapses may be the workhorse of the neuronal computations that underlie probabilistic reasoning. We built a neural circuit model for probabilistic inference in which information provided by different sensory cues must be integrated and the predictive powers of individual cues about an outcome are deduced through experience. We found that bounded synapses naturally compute, through reward-dependent plasticity, the posterior probability that a choice alternative is correct given that a cue is presented. Furthermore, a decision circuit endowed with such synapses makes choices on the basis of the summed log posterior odds and performs near-optimal cue combination. The model was validated by reproducing salient observations of, and provides insights into, a monkey experiment using a categorization task. Our model thus suggests a biophysical instantiation of the Bayesian decision rule, while predicting important deviations from it similar to the 'base-rate neglect' observed in human studies when alternatives have unequal prior probabilities.",
author = "Alireza Soltani and Xiao-Jing Wang",
year = "2010",
month = "1",
day = "17",
doi = "10.1038/nn.2450",
language = "English (US)",
volume = "13",
pages = "112--119",
journal = "Nature Neuroscience",
issn = "1097-6256",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - Synaptic computation underlying probabilistic inference

AU - Soltani, Alireza

AU - Wang, Xiao-Jing

PY - 2010/1/17

Y1 - 2010/1/17

N2 - We propose that synapses may be the workhorse of the neuronal computations that underlie probabilistic reasoning. We built a neural circuit model for probabilistic inference in which information provided by different sensory cues must be integrated and the predictive powers of individual cues about an outcome are deduced through experience. We found that bounded synapses naturally compute, through reward-dependent plasticity, the posterior probability that a choice alternative is correct given that a cue is presented. Furthermore, a decision circuit endowed with such synapses makes choices on the basis of the summed log posterior odds and performs near-optimal cue combination. The model was validated by reproducing salient observations of, and provides insights into, a monkey experiment using a categorization task. Our model thus suggests a biophysical instantiation of the Bayesian decision rule, while predicting important deviations from it similar to the 'base-rate neglect' observed in human studies when alternatives have unequal prior probabilities.

AB - We propose that synapses may be the workhorse of the neuronal computations that underlie probabilistic reasoning. We built a neural circuit model for probabilistic inference in which information provided by different sensory cues must be integrated and the predictive powers of individual cues about an outcome are deduced through experience. We found that bounded synapses naturally compute, through reward-dependent plasticity, the posterior probability that a choice alternative is correct given that a cue is presented. Furthermore, a decision circuit endowed with such synapses makes choices on the basis of the summed log posterior odds and performs near-optimal cue combination. The model was validated by reproducing salient observations of, and provides insights into, a monkey experiment using a categorization task. Our model thus suggests a biophysical instantiation of the Bayesian decision rule, while predicting important deviations from it similar to the 'base-rate neglect' observed in human studies when alternatives have unequal prior probabilities.

UR - http://www.scopus.com/inward/record.url?scp=73949142431&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=73949142431&partnerID=8YFLogxK

U2 - 10.1038/nn.2450

DO - 10.1038/nn.2450

M3 - Article

C2 - 20010823

AN - SCOPUS:73949142431

VL - 13

SP - 112

EP - 119

JO - Nature Neuroscience

JF - Nature Neuroscience

SN - 1097-6256

IS - 1

ER -