Network Structure within the Cerebellar Input Layer Enables Lossless Sparse Encoding

Guy Billings,Eugenio Piasini,Andrea Lőrincz,Zoltan Nusser,R Angus Silver

doi:10.1016/j.neuron.2014.07.020

Abstract

SummaryThe synaptic connectivity within neuronal networks is thought to determine the information processing they perform, yet network structure-function relationships remain poorly understood. By combining quantitative anatomy of the cerebellar input layer and information theoretic analysis of network models, we investigated how synaptic connectivity affects information transmission and processing. Simplified binary models revealed that the synaptic connectivity within feedforward networks determines the trade-off between information transmission and sparse encoding. Networks with few synaptic connections per neuron and network-activity-dependent threshold were optimal for lossless sparse encoding over the widest range of input activities. Biologically detailed spiking network models with experimentally constrained synaptic conductances and inhibition confirmed our analytical predictions. Our results establish that the synaptic connectivity within the cerebellar input layer enables efficient lossless sparse encoding. Moreover, they provide a functional explanation for why granule cells have approximately four dendrites, a feature that has been evolutionarily conserved since the appearance of fish.

Highlights

Different regions of the brain exhibit distinct anatomical structures, cell morphologies, and synaptic connectivities and perform specific computational tasks
Our results show that the synaptic connectivity within the cerebellar input layer, where GCs receive an average of approximately four excitatory mossy fiber (MF) inputs, is well suited for performing sparse encoding without loss of information
Trade-Off between Information Transmission and Sparsification in Uniform Binary Network Models with Fixed GC Threshold To investigate further how network connectivity affects information transmission and sparsification, we examined the case of a relatively high fixed threshold, since GCs typically require activation of three of their four MF inputs to fire (Jorntell and Ekerot, 2006; Schwartz et al, 2012) and spike threshold is dominated by the presence of a large tonic inhibitory conductance (Brickley et al, 1996; Duguid et al, 2012)

Summary

Introduction

Different regions of the brain exhibit distinct anatomical structures, cell morphologies, and synaptic connectivities and perform specific computational tasks. Linking the structure to function (Honey et al, 2007) or dysfunction (Dyhrfjeld-Johnsen et al, 2007) has proved difficult, because the synaptic connectivity, neuronal properties, and the computations performed are usually poorly defined. In pattern generator circuits within the spinal cord, distinct neuronal subtypes compute different gaits during locomotion (Talpalar et al, 2013). Despite these advances, the contribution that synaptic connectivity makes to information processing remains unclear in most brain regions

Results

Discussion

Conclusion