Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning

Average human behavior in cue combination tasks is well predicted by Bayesian inference models. As this capability is acquired over developmental timescales, the question arises, how it is learned. Here we investigated whether reward dependent learning, that is well established at the computational,...

Full description

Bibliographic Details
Main Authors:	Weisswange, Thomas H., Rothkopf, Constantin A., Rodemann, Tobias, Triesch, Jochen
Format:	Online
Language:	English
Published:	Public Library of Science 2011
Online Access:	https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3130032/

id	pubmed-3130032
recordtype	oai_dc
spelling	pubmed-31300322011-07-12 Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning Weisswange, Thomas H. Rothkopf, Constantin A. Rodemann, Tobias Triesch, Jochen Research Article Average human behavior in cue combination tasks is well predicted by Bayesian inference models. As this capability is acquired over developmental timescales, the question arises, how it is learned. Here we investigated whether reward dependent learning, that is well established at the computational, behavioral, and neuronal levels, could contribute to this development. It is shown that a model free reinforcement learning algorithm can indeed learn to do cue integration, i.e. weight uncertain cues according to their respective reliabilities and even do so if reliabilities are changing. We also consider the case of causal inference where multimodal signals can originate from one or multiple separate objects and should not always be integrated. In this case, the learner is shown to develop a behavior that is closest to Bayesian model averaging. We conclude that reward mediated learning could be a driving force for the development of cue integration and causal inference. Public Library of Science 2011-07-05 /pmc/articles/PMC3130032/ /pubmed/21750717 http://dx.doi.org/10.1371/journal.pone.0021575 Text en Weisswange et al. http://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
repository_type	Open Access Journal
institution_category	Foreign Institution
institution	US National Center for Biotechnology Information
building	NCBI PubMed
collection	Online Access
language	English
format	Online
author	Weisswange, Thomas H. Rothkopf, Constantin A. Rodemann, Tobias Triesch, Jochen
spellingShingle	Weisswange, Thomas H. Rothkopf, Constantin A. Rodemann, Tobias Triesch, Jochen Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
author_facet	Weisswange, Thomas H. Rothkopf, Constantin A. Rodemann, Tobias Triesch, Jochen
author_sort	Weisswange, Thomas H.
title	Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
title_short	Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
title_full	Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
title_fullStr	Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
title_full_unstemmed	Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning
title_sort	bayesian cue integration as a developmental outcome of reward mediated learning
description	Average human behavior in cue combination tasks is well predicted by Bayesian inference models. As this capability is acquired over developmental timescales, the question arises, how it is learned. Here we investigated whether reward dependent learning, that is well established at the computational, behavioral, and neuronal levels, could contribute to this development. It is shown that a model free reinforcement learning algorithm can indeed learn to do cue integration, i.e. weight uncertain cues according to their respective reliabilities and even do so if reliabilities are changing. We also consider the case of causal inference where multimodal signals can originate from one or multiple separate objects and should not always be integrated. In this case, the learner is shown to develop a behavior that is closest to Bayesian model averaging. We conclude that reward mediated learning could be a driving force for the development of cue integration and causal inference.
publisher	Public Library of Science
publishDate	2011
url	https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3130032/
_version_	1611463828608909312

Bayesian Cue Integration as a Developmental Outcome of Reward Mediated Learning

Similar Items