Condensed Matter > Disordered Systems and Neural Networks
[Submitted on 27 Aug 2015 (v1), revised 7 Jan 2016 (this version, v3), latest version 27 Apr 2017 (v4)]
Title:Continuous parameter working memory in a balanced chaotic neural network
View PDFAbstract:Working memory, the ability to maintain information over time scales greater than those characterizing single neurons, is essential to many brain functions. It remains unclear whether neural networks in the balanced state, an important model for activity in the cortex, can support a continuum of stable states that would make it possible to store a continuous variable in working memory while also accounting for the stochastic behavior of single neurons. Here we propose a simple neural architecture that achieves this goal. We show analytically that in the limit of an infinite network a continuous parameter can be stored indefinitely on a continuum of balanced states. For finite networks we calculate the diffusivity along the attractor driven by the chaotic noise in the network, and show that it is inversely proportional to the system size. Thus, for large enough (but realistic) neural population sizes, and with suitable tuning of the network connections, it is possible to maintain continuous parameter values over time scales larger by several orders of magnitude than the single neuron time scale.
Submission history
From: Yoram Burak [view email][v1] Thu, 27 Aug 2015 17:24:13 UTC (2,096 KB)
[v2] Tue, 1 Sep 2015 14:25:08 UTC (4,020 KB)
[v3] Thu, 7 Jan 2016 09:15:14 UTC (3,767 KB)
[v4] Thu, 27 Apr 2017 15:39:27 UTC (1,539 KB)
Current browse context:
cond-mat.dis-nn
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
IArxiv Recommender
(What is IArxiv?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.