Computation is concentrated in rich clubs of local cortical networks.

John M Beggs,Samantha P Faber,Ehren L Newman,Nicholas M Timme

doi:10.1162/netn_a_00069

Abstract

To understand how neural circuits process information, it is essential to identify the relationship between computation and circuit organization. Rich clubs, highly interconnected sets of neurons, are known to propagate a disproportionate amount of information within cortical circuits. Here, we test the hypothesis that rich clubs also perform a disproportionate amount of computation. To do so, we recorded the spiking activity of on average ∼300 well-isolated individual neurons from organotypic cortical cultures. We then constructed weighted, directed networks reflecting the effective connectivity between the neurons. For each neuron, we quantified the amount of computation it performed based on its inputs. We found that rich-club neurons compute ∼160% more information than neurons outside of the rich club. The amount of computation performed in the rich club was proportional to the amount of information propagation by the same neurons. This suggests that in these circuits, information propagation drives computation. In total, our findings indicate that rich-club organization in effective cortical circuits supports not only information propagation but also neural computation.

Highlights

The idea that neurons propagate information and that downstream neurons integrate this information via neural computation is foundational to our understanding of how the brain processes, and responds to, the world
Our analysis of neural computation was predicated on the idea that nonlinear integration of multiple inputs implements a form of computation
partial information decomposition approach (PID) is currently the only method available that is capable of quantifying the computation that occurs in the multiway interactions of neuronal triads

Summary

Introduction

The idea that neurons propagate information and that downstream neurons integrate this information via neural computation is foundational to our understanding of how the brain processes, and responds to, the world. The determinants of such computations remain largely unknown. Advances in data acquisition methods, offering increasingly comprehensive recordings of the activation dynamics that play out atop of neural circuits, together with advances in data analytics, make it possible to empirically study the determinants of neural. Information: The reduction in uncertainty, typically measured in bits. Computation: The process of integrating multiple sources of information to produce an output. Information propagation: The transfer of unmodified information from one spiking neuron to another. Rich club: A set of neurons with strong connections that connect to each other more than expected by chance

Objectives

Methods

Results

Discussion

Conclusion