Temporal chunking as a mechanism for unsupervised learning of task-sets
Depending on environmental demands, humans can learn and exploit multiple concurrent sets of stimulus-response associations. Mechanisms underlying the learning of such task-sets remain unknown. Here we investigate the hypothesis that task-set learning relies on unsupervised chunking of stimulus-resp...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
eLife Sciences Publications Ltd
2020-03-01
|
Series: | eLife |
Subjects: | |
Online Access: | https://elifesciences.org/articles/50469 |
id |
doaj-ac336e9a6ded411897f56152ef5c1a42 |
---|---|
record_format |
Article |
spelling |
doaj-ac336e9a6ded411897f56152ef5c1a422021-05-05T20:53:53ZengeLife Sciences Publications LtdeLife2050-084X2020-03-01910.7554/eLife.50469Temporal chunking as a mechanism for unsupervised learning of task-setsFlora Bouchacourt0https://orcid.org/0000-0002-8893-0143Stefano Palminteri1https://orcid.org/0000-0001-5768-6646Etienne Koechlin2Srdjan Ostojic3https://orcid.org/0000-0002-7473-1223Laboratoire de Neurosciences Cognitives et Computationnelles, Institut National de la Sante et de la Recherche Medicale, Paris, France; Departement d’Etudes Cognitives, Ecole Normale Superieure, Paris, FranceLaboratoire de Neurosciences Cognitives et Computationnelles, Institut National de la Sante et de la Recherche Medicale, Paris, France; Departement d’Etudes Cognitives, Ecole Normale Superieure, Paris, France; Institut d’Etudes de la Cognition, Universite de Recherche Paris Sciences et Lettres, Paris, FranceLaboratoire de Neurosciences Cognitives et Computationnelles, Institut National de la Sante et de la Recherche Medicale, Paris, France; Departement d’Etudes Cognitives, Ecole Normale Superieure, Paris, FranceLaboratoire de Neurosciences Cognitives et Computationnelles, Institut National de la Sante et de la Recherche Medicale, Paris, France; Departement d’Etudes Cognitives, Ecole Normale Superieure, Paris, France; Institut d’Etudes de la Cognition, Universite de Recherche Paris Sciences et Lettres, Paris, FranceDepending on environmental demands, humans can learn and exploit multiple concurrent sets of stimulus-response associations. Mechanisms underlying the learning of such task-sets remain unknown. Here we investigate the hypothesis that task-set learning relies on unsupervised chunking of stimulus-response associations that occur in temporal proximity. We examine behavioral and neural data from a task-set learning experiment using a network model. We first show that task-set learning can be achieved provided the timescale of chunking is slower than the timescale of stimulus-response learning. Fitting the model to behavioral data on a subject-by-subject basis confirmed this expectation and led to specific predictions linking chunking and task-set retrieval that were borne out by behavioral performance and reaction times. Comparing the model activity with BOLD signal allowed us to identify neural correlates of task-set retrieval in a functional network involving ventral and dorsal prefrontal cortex, with the dorsal system preferentially engaged when retrievals are used to improve performance.https://elifesciences.org/articles/50469computational neurosciencecognitive neuroscienceneural networks |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Flora Bouchacourt Stefano Palminteri Etienne Koechlin Srdjan Ostojic |
spellingShingle |
Flora Bouchacourt Stefano Palminteri Etienne Koechlin Srdjan Ostojic Temporal chunking as a mechanism for unsupervised learning of task-sets eLife computational neuroscience cognitive neuroscience neural networks |
author_facet |
Flora Bouchacourt Stefano Palminteri Etienne Koechlin Srdjan Ostojic |
author_sort |
Flora Bouchacourt |
title |
Temporal chunking as a mechanism for unsupervised learning of task-sets |
title_short |
Temporal chunking as a mechanism for unsupervised learning of task-sets |
title_full |
Temporal chunking as a mechanism for unsupervised learning of task-sets |
title_fullStr |
Temporal chunking as a mechanism for unsupervised learning of task-sets |
title_full_unstemmed |
Temporal chunking as a mechanism for unsupervised learning of task-sets |
title_sort |
temporal chunking as a mechanism for unsupervised learning of task-sets |
publisher |
eLife Sciences Publications Ltd |
series |
eLife |
issn |
2050-084X |
publishDate |
2020-03-01 |
description |
Depending on environmental demands, humans can learn and exploit multiple concurrent sets of stimulus-response associations. Mechanisms underlying the learning of such task-sets remain unknown. Here we investigate the hypothesis that task-set learning relies on unsupervised chunking of stimulus-response associations that occur in temporal proximity. We examine behavioral and neural data from a task-set learning experiment using a network model. We first show that task-set learning can be achieved provided the timescale of chunking is slower than the timescale of stimulus-response learning. Fitting the model to behavioral data on a subject-by-subject basis confirmed this expectation and led to specific predictions linking chunking and task-set retrieval that were borne out by behavioral performance and reaction times. Comparing the model activity with BOLD signal allowed us to identify neural correlates of task-set retrieval in a functional network involving ventral and dorsal prefrontal cortex, with the dorsal system preferentially engaged when retrievals are used to improve performance. |
topic |
computational neuroscience cognitive neuroscience neural networks |
url |
https://elifesciences.org/articles/50469 |
work_keys_str_mv |
AT florabouchacourt temporalchunkingasamechanismforunsupervisedlearningoftasksets AT stefanopalminteri temporalchunkingasamechanismforunsupervisedlearningoftasksets AT etiennekoechlin temporalchunkingasamechanismforunsupervisedlearningoftasksets AT srdjanostojic temporalchunkingasamechanismforunsupervisedlearningoftasksets |
_version_ |
1721458526982766592 |