Work

Principles of inter-areal connections of the macaque cortex

Year: 2010

Type: conference-paper

Abstract: The operation of real world networks is largely determined by their weighted and spatial characteristics. Surprisingly little is known about these features in cortex. We generated in macaque, a consistent database of inter-areal connections comprising projection densities (link weights) and physical lengths. Contrary to previous assumptions, the cortical connection matrix is dense (66%) and therefore, not a small-world graph. Link weights are both highly specific and heterogeneous and we show that it is these properties that characterize the network. The embedding of this weighted network is governed by a distance rule that predicts both its binary features as well as the global and local communication efficiencies. Analysis of the efficiency of this weighted network suggests that small changes in global communication efficiency are offset by large changes in local efficiency. These findings indicate a weight-based hierarchical layering in cortical architecture and processing.

Source: HAL