English
 
Help Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
  Examining network dynamics under inhibitory rewiring

Patzlaff, N., Vinogradov, O., Giannakakis, E., & Levina, A. (2021). Examining network dynamics under inhibitory rewiring. Poster presented at Bernstein Conference 2021. doi:10.12751/nncn.bc2021.p201.

Item is

Files

show Files

Creators

show
hide
 Creators:
Patzlaff, N, Author
Vinogradov, O1, 2, Author           
Giannakakis, E1, 2, Author           
Levina, A1, 2, Author           
Affiliations:
1Department of Computational Neuroscience, Max Planck Institute for Biological Cybernetics, Max Planck Society, ou_3017468              
2Max Planck Institute for Biological Cybernetics, Max Planck Society, ou_1497794              

Content

show
hide
Free keywords: -
 Abstract: Inhibition is crucial for the stability of network dynamics. In particular, rewiring inhibitory connectivity was shown [1] to have a drastic effect on firing rate dynamics (as measured by correlations of the population rate vector before and after rewiring (Fig. 1B)). At the same time, the effect of rewiring the more common excitatory cells is less pronounced. Part of this disproportionality has been attributed to the wider distribution of inhibitory firing rates [1], however, this hypothesis was not tested explicitly. We use a simple recurrent E/I network to test factors underlying the stability of firing rates to rewiring of E-weights and its fragility to I-rewiring. We examine multiple factors that have been shown to affect the firing behaviour of neuronal networks and thus can influence the impact of inhibitory connectivity. We simulate a network of LIF neurons with various E/I-ratios and connection weights drawn from log-normal distributions (Fig. 1A). We keep variances the same for all weights, with means adjusted depending on E/I-ratio to keep the network in the asynchronous irregular (AI) regime. We find that the disproportionate effect of inhibition is maintained under a variety of configurations and is therefore effectively independent of the E/I-ratio (Fig. 1 C&D), rate and weight distributions as long as the AI state is maintained. Additionally, changes in the relative firing rates of the excitatory and inhibitory populations seem to also not affect the relative impact of inhibitory connectivity that may only be reduced by transitioning to a non-AI network state. As a next step, we further test how specific parameter changes affect rewiring while keeping network behaviour constant. For that, we train a masked flow as a conditional density estimator [2] to return posterior distributions of parameters given a set of summary statistics. So far we conclude that the disproportionate effect of inhibition on network dynamics can not be completely explained by the relative variance of excitatory and inhibitory firing rate distributions, but that other factors should be relevant.

Details

show
hide
Language(s):
 Dates: 2021-09
 Publication Status: Published online
 Pages: -
 Publishing info: -
 Table of Contents: -
 Rev. Type: -
 Identifiers: DOI: 10.12751/nncn.bc2021.p201
 Degree: -

Event

show
hide
Title: Bernstein Conference 2021
Place of Event: -
Start-/End Date: 2021-09-21 - 2021-09-24

Legal Case

show

Project information

show

Source 1

show
hide
Title: Bernstein Conference 2021
Source Genre: Proceedings
 Creator(s):
Affiliations:
Publ. Info: -
Pages: - Volume / Issue: - Sequence Number: P 201 Start / End Page: - Identifier: -