Inductive biases of neural network modularity in spatial navigation

Simple item page
dc.citation.articleNumbereadk1256
dc.citation.issueNumber29
dc.citation.journalTitleScience Advances
dc.citation.volumeNumber10
dc.contributor.authorZhang, Ruiyi
dc.contributor.authorPitkow, Xaq
dc.contributor.authorAngelaki, Dora E.
dc.date.accessioned2024-08-09T16:25:26Z
dc.date.available2024-08-09T16:25:26Z
dc.date.issued2024
dc.description.abstractThe brain may have evolved a modular architecture for daily tasks, with circuits featuring functionally specialized modules that match the task structure. We hypothesize that this architecture enables better learning and generalization than architectures with less specialized modules. To test this, we trained reinforcement learning agents with various neural architectures on a naturalistic navigation task. We found that the modular agent, with an architecture that segregates computations of state representation, value, and action into specialized modules, achieved better learning and generalization. Its learned state representation combines prediction and observation, weighted by their relative uncertainty, akin to recursive Bayesian estimation. This agent’s behavior also resembles macaques’ behavior more closely. Our results shed light on the possible rationale for the brain’s modularity and suggest that artificial systems can use this insight from neuroscience to improve learning and generalization in natural tasks.
dc.identifier.citationZhang, R., Pitkow, X., & Angelaki, D. E. (2024). Inductive biases of neural network modularity in spatial navigation. Science Advances, 10(29), eadk1256. https://doi.org/10.1126/sciadv.adk1256
dc.identifier.digitalsciadv-adk1256
dc.identifier.doihttps://doi.org/10.1126/sciadv.adk1256
dc.identifier.urihttps://hdl.handle.net/1911/117647
dc.language.isoeng
dc.publisherAAAS
dc.rightsExcept where otherwise noted, this work is licensed under a Creative Commons Attribution (CC BY) license.  Permission to reuse, publish, or reproduce the work beyond the terms of the license or beyond the bounds of fair use or other exemptions to copyright law must be obtained from the copyright holder.
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.titleInductive biases of neural network modularity in spatial navigation
dc.typeJournal article
dc.type.dcmiText
dc.type.publicationpublisher version
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
sciadv-adk1256.pdf
Size:
2.16 MB
Format:
Adobe Portable Document Format
Download
Collections
Faculty Publications
ECE Publications