Quantitative Biology > Neurons and Cognition
[Submitted on 11 Jul 2013 (v1), revised 12 Jul 2013 (this version, v2), latest version 15 Jan 2014 (v4)]
Title:The sign rule and beyond: Boundary effects, flexibility, and optimal noise correlations in neural population codes
View PDFAbstract:Over repeat presentations of the same stimulus, sensory neurons show variable responses. This noise exhibited by pairs of cells is typically correlated, and a question with rich history in neuroscience is how these noise correlations impact the population's ability to encode the stimulus. A complex and interesting picture has emerged from prior work, with different effects seen in highly structured vs. heterogeneous neural populations.
Here, we consider a very general setting for population coding, and cast the problem in terms of optimizing information as a function of noise correlations, with all other aspects of the problem - like neural tuning curves, etc. - held fixed. That work yields some unifying insights into the role of noise correlations. These are stated as several theorems, and illustrated in numerical examples of neurons with diverse tuning curves. Our main contributions are as follows.
(1) We generalize previous results to prove a sign rule (SR) - if noise correlations between pairs of neurons have opposite signs vs. their signal correlations, then coding performance will improve vs. the independent case. This holds for three different metrics of coding performance, and for arbitrary tuning curves and levels of heterogeneity. (2) We prove that the optimal correlation structures must lie on boundaries of the possible set of noise correlations. Moreover, these boundary solutions are very numerous and often violate the SR. Thus, noise correlations that violate the SR - which are often observed experimentally - can yield better coding performance than can ones that obey it. (3) We provide a set of necessary and sufficient conditions, under which the coding performance (in the presence of noise) will be as good as it would be if there were no noise present at all.
Submission history
From: Yu Hu [view email][v1] Thu, 11 Jul 2013 19:52:20 UTC (1,216 KB)
[v2] Fri, 12 Jul 2013 19:49:11 UTC (1,217 KB)
[v3] Thu, 22 Aug 2013 18:52:35 UTC (3,810 KB)
[v4] Wed, 15 Jan 2014 23:46:25 UTC (3,799 KB)
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?)
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.