Optimal features for auditory categorization

Shi Tong Liu, Pilar Montes-Lourido, Xiaoqin Wang, Srivatsun Sadagopan

Research output: Contribution to journalArticle

Abstract

Humans and vocal animals use vocalizations to communicate with members of their species. A necessary function of auditory perception is to generalize across the high variability inherent in vocalization production and classify them into behaviorally distinct categories (‘words’ or ‘call types’). Here, we demonstrate that detecting mid-level features in calls achieves production-invariant classification. Starting from randomly chosen marmoset call features, we use a greedy search algorithm to determine the most informative and least redundant features necessary for call classification. High classification performance is achieved using only 10–20 features per call type. Predictions of tuning properties of putative feature-selective neurons accurately match some observed auditory cortical responses. This feature-based approach also succeeds for call categorization in other species, and for other complex classification tasks such as caller identification. Our results suggest that high-level neural representations of sounds are based on task-dependent features optimized for specific computational goals.

Original languageEnglish (US)
Article number1302
JournalNature communications
Volume10
Issue number1
DOIs
StatePublished - Dec 1 2019

Fingerprint

Animal Vocalization
auditory perception
Auditory Perception
Callithrix
neurons
Neurons
animals
Animals
Tuning
tuning
Acoustic waves
acoustics
predictions
Identification (Psychology)

ASJC Scopus subject areas

  • Chemistry(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Physics and Astronomy(all)

Cite this

Optimal features for auditory categorization. / Liu, Shi Tong; Montes-Lourido, Pilar; Wang, Xiaoqin; Sadagopan, Srivatsun.

In: Nature communications, Vol. 10, No. 1, 1302, 01.12.2019.

Research output: Contribution to journalArticle

Liu, Shi Tong ; Montes-Lourido, Pilar ; Wang, Xiaoqin ; Sadagopan, Srivatsun. / Optimal features for auditory categorization. In: Nature communications. 2019 ; Vol. 10, No. 1.
@article{f271705c4e0d4905941218c7fc83306d,
title = "Optimal features for auditory categorization",
abstract = "Humans and vocal animals use vocalizations to communicate with members of their species. A necessary function of auditory perception is to generalize across the high variability inherent in vocalization production and classify them into behaviorally distinct categories (‘words’ or ‘call types’). Here, we demonstrate that detecting mid-level features in calls achieves production-invariant classification. Starting from randomly chosen marmoset call features, we use a greedy search algorithm to determine the most informative and least redundant features necessary for call classification. High classification performance is achieved using only 10–20 features per call type. Predictions of tuning properties of putative feature-selective neurons accurately match some observed auditory cortical responses. This feature-based approach also succeeds for call categorization in other species, and for other complex classification tasks such as caller identification. Our results suggest that high-level neural representations of sounds are based on task-dependent features optimized for specific computational goals.",
author = "Liu, {Shi Tong} and Pilar Montes-Lourido and Xiaoqin Wang and Srivatsun Sadagopan",
year = "2019",
month = "12",
day = "1",
doi = "10.1038/s41467-019-09115-y",
language = "English (US)",
volume = "10",
journal = "Nature Communications",
issn = "2041-1723",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - Optimal features for auditory categorization

AU - Liu, Shi Tong

AU - Montes-Lourido, Pilar

AU - Wang, Xiaoqin

AU - Sadagopan, Srivatsun

PY - 2019/12/1

Y1 - 2019/12/1

N2 - Humans and vocal animals use vocalizations to communicate with members of their species. A necessary function of auditory perception is to generalize across the high variability inherent in vocalization production and classify them into behaviorally distinct categories (‘words’ or ‘call types’). Here, we demonstrate that detecting mid-level features in calls achieves production-invariant classification. Starting from randomly chosen marmoset call features, we use a greedy search algorithm to determine the most informative and least redundant features necessary for call classification. High classification performance is achieved using only 10–20 features per call type. Predictions of tuning properties of putative feature-selective neurons accurately match some observed auditory cortical responses. This feature-based approach also succeeds for call categorization in other species, and for other complex classification tasks such as caller identification. Our results suggest that high-level neural representations of sounds are based on task-dependent features optimized for specific computational goals.

AB - Humans and vocal animals use vocalizations to communicate with members of their species. A necessary function of auditory perception is to generalize across the high variability inherent in vocalization production and classify them into behaviorally distinct categories (‘words’ or ‘call types’). Here, we demonstrate that detecting mid-level features in calls achieves production-invariant classification. Starting from randomly chosen marmoset call features, we use a greedy search algorithm to determine the most informative and least redundant features necessary for call classification. High classification performance is achieved using only 10–20 features per call type. Predictions of tuning properties of putative feature-selective neurons accurately match some observed auditory cortical responses. This feature-based approach also succeeds for call categorization in other species, and for other complex classification tasks such as caller identification. Our results suggest that high-level neural representations of sounds are based on task-dependent features optimized for specific computational goals.

UR - http://www.scopus.com/inward/record.url?scp=85063322729&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063322729&partnerID=8YFLogxK

U2 - 10.1038/s41467-019-09115-y

DO - 10.1038/s41467-019-09115-y

M3 - Article

C2 - 30899018

AN - SCOPUS:85063322729

VL - 10

JO - Nature Communications

JF - Nature Communications

SN - 2041-1723

IS - 1

M1 - 1302

ER -