Network attributes for segregation and integration in the human brain

Network studies of large-scale brain connectivity have begun to reveal attributes that promote the segregation and integration of neural information: communities and hubs. Network communities are sets of regions that are strongly interconnected among each other while connections between members of different communities are less dense. The clustered connectivity of network communities supports functional segregation and specialization. Network hubs link communities to one another and ensure efficient communication and information integration. This review surveys a number of recent reports on network communities and hubs, and their role in integrative processes. An emerging focus is the shifting balance between segregation and integration over time, which manifest in continuously changing patterns of functional interactions between regions, circuits and systems.

Introduction

Recent years have seen a sharp increase in empirical and theoretical studies of networks as models of complex systems. In neuroscience, the rising interest in brain networks is driven by the increasing availability of network data on the structure and function of neural systems. Such networks or graphs, described as collections of nodes (neurons, regions) and edges (connections, pathways) can be analyzed with a wide array of quantitative tools and methods (Figure 1) [1, 2, 3, 4, 5, 6, 7]. Importantly, network science not only provides intuitive and analytically powerful approaches for data analysis and modeling, it also offers a comprehensive theoretical framework for understanding the biological basis of brain function [8]. This framework bridges and unifies the domains of neuroanatomy (‘structural connectivity’ [9]) and brain dynamics (‘functional and effective connectivity’ [10]) by linking neuronal operations (measured empirically or generated computationally) to an underlying anatomical substrate.

This review article surveys a selection of recent studies on large-scale brain networks, mostly obtained from noninvasive imaging of the human brain. What these studies have in common is that they use network approaches to gain insight into the basis of integrative brain function. Structural connections are fundamental in this regard because they allow neural elements to coordinate their activity into coherent dynamic states that support cognition and behavior. To achieve such coherent dynamics, structural networks shape the flow of information between local regions of the brain to accomplish two distinct goals (Figure 1): firstly they promote functional segregation by forming local network communities that are intrinsically densely connected and strongly coupled; and secondly they promote functional integration by enabling global communication between communities through network hubs. The balance between segregation and integration is essential for the operation of distributed networks underlying cognitive function [11, 12]. The remainder of this review will survey recent studies that have identified network architectures and mechanisms that promote segregation, integration, and their dynamic interplay.