Optimization of signal transmission between neural populations by stimulation of driver nodes

Context and relevance. Synchronized activity is critical in brain neural networks, yet the precise mechanisms that ensure coherence of activity between different brain parts remain unclear. Objective. To determine how the choice of driver nodes (according to different centrality metrics) affects the propagation of spiking activity between two sparsely connected clusters. Hypothesis. The selection of control nodes in the first cluster improves the propagation of spike activity to the second cluster compared to a random selection of neurons. Methods and Materials. In a two-cluster network modeled by a stochastic block model, a fraction of neurons in the first cluster (10-20%) were subjected to external stimulation. Driver nodes were selected randomly or by centrality: betweenness, closeness, degree, eigenvector, harmonic, and percolation. Results. When stimulating the neurons chosen on the base of various centrality measures, the average firing rate in the second cluster increased proportionally to the number of driver nodes and synchronized with the first cluster. Conclusions. Network topology and driver node selection metrics determine the efficiency and robustness of inter-cluster synchronization, which is important for neuromodulation and the development of bio-inspired computing systems.