Effect of Columnar Neural Grouping on Network Synchronization

Abstract

Connectivity in the brain has long been explored on varying scales: from connectivity of large regions down to groups of only a few neurons. In this work we explore how a connectivity scheme inspired by columnar organization in the neocortex effects the synchronization of a system of neurons. Neurons are grouped based on (x,y) positions in a grid to form columnar groups, and a control system is grouped independent of position. These neurons are connected with a bias towards connections within their groups, based on bias parameter, δ. This scheme leads to a small-world network containing highly connected groups as well as inter-group connections. Systems were created with purely excitatory neurons as well as systems of 80% excitatory neurons and 20% inhibitory neurons. Simulations are run with external stimulation to each Hodgkin-Huxley type neuron, and distance-dependent delays are included in the propagation of each action potential. We find that initially the connectivity must reach a critical point in order for the system to synchronize. Above that point, the randomly grouped control system becomes less synchronous (measured by SPIKE-distance) than the system grouped in a columnar structure. This difference is significant in a particular region of our parameter phase space at very low signal propagation speeds. Additional work has been done to further characterize the dynamics of these systems including the frequency of global spiking events.