This title appears in the Scientific Report :
2018
Inhibitory assemblies play an important role in cortical attractor networks
Inhibitory assemblies play an important role in cortical attractor networks
Balanced networks of inhibitory and excitatory neurons with homogeneously random recurrent connectivity are often employed to model local cortical circuits. The balanced random network model exhibits irregular and asynchronous spiking activity similar to that observed in vivo. A recent series of stu...
Saved in:
Personal Name(s): | Rostami, Vahid (Corresponding author) |
---|---|
Rost, Thomas / van Albada, Sacha / Nawrot, Martin | |
Contributing Institute: |
Jara-Institut Brain structure-function relationships; INM-10 Computational and Systems Neuroscience; IAS-6 Computational and Systems Neuroscience; INM-6 |
Imprint: |
2018
|
Conference: | 7th International caesar Conference, Bonn (Germany), 2018-06-05 - 2018-06-06 |
Document Type: |
Abstract |
Research Program: |
Connectivity and Activity |
Publikationsportal JuSER |
Balanced networks of inhibitory and excitatory neurons with homogeneously random recurrent connectivity are often employed to model local cortical circuits. The balanced random network model exhibits irregular and asynchronous spiking activity similar to that observed in vivo. A recent series of studies [1-3] has extended the balanced random network model to incorporate clusters of strongly interconnected excitatory neurons, with no modularity in the inhibitory population. This clustered topology demonstrates a functionally desired multistability where different clusters become spontaneously activated and inactivated. The model captures a realistic high firing variability of single neurons and the reduction in trial-to-trial variability during stimulation of clusters as observed experimentally.We recently showed that, despite the multistability and trial-to-trial variability that emerge in the clustered excitatory network, this topology leads to widely separated firing rate states of single neurons and tends quickly towards firing rate saturation, which is inconsistent with experimental observations. To overcome this problem we introduced clusters of inhibitory neurons which are coupled to each excitatory cluster [4]. This connectivity scheme is not directly supported by experimental findings. However, recent anatomical and physiological studies point to increased local inhibitory connectivities and possible inhibitory clustering through connection strengths [5-7].Here we model different architectures of inhibitory circuits, based on these recent experimental studies, and investigate the role of inhibitory clusters on the multistability and trial-to-trial variability of the spiking network when excitatory clusters have strengthened connections with different portions of the inhibitory population. Our model can be reduced to the case of exclusively excitatory clusters [2], or to a one-to-one correspondence of inhibitory and excitatory clusters [4], but we explore all different architectures in between these extreme cases. Such intermediate scenarios are more consistent with recent experimental observations.We find that inhibitory clustering is necessary to achieve realistic spiking activity under stimulation in terms of a biologically realistic firing rate, spiking regularity, and trial-to-trial spike count variability. Inhibitory clustering achieves the desired attractor dynamics over a wide range of network parameters and thus makes networks robust against parameter fluctuations due to homeostasis or neuromodulation. Remarkably, when the stimulus is weak, without clustering of inhibitory neurons, the spiking network model fails to capture the reduction of trial-to-trial variability during stimulation.AcknowledgmentsThis work is supported by the German Science Foundation under the Institutional Strategy of the University of Cologne within the German Excellence Initiative (DFG-ZUK 81/1).References1.Deco, Hugues (2012) PLOS Comput. Biol., 8(3): e1002395.2.Litwin-Kumar, Doiron (2012) Nat. Neurosci., 15(11), 1498–1505.3.Mazzucato, Fontanini, La Camera (2015) J. Neurosci. 35(21):8214–8231.4.Rost, Deger, Nawrot (2017) Biol. Cybern., doi: 10.1007/s00422-017-0737-7.5.Xue, Atallah, Scanziani (2014) Nature, 511(7511), 596–600.6.Lee, Marchionni, Bezaire, et al., (2014) Neuron 82, 1129–1144.7.Morishima, Kobayashi, Kato, et al., (2017) Cereb. Cortex 27, 5846–5857. |