Delay selection by spike-timing-dependent plasticity shapes efficient networks for signal transmission

Abstract

Information processing in the brain relies on efficient communication between different brain regions. Brain oscillations can control signal transmission in brain networks by modulating the timing and excitability of sender and receiver areas. For effective transmission, signals should arrive at target areas when their excitability is maximized. For reciprocally connected neural populations, this mechanism works if the transmission delay matches the period of their evoked oscillation. However, the mechanisms underlying such development of the connections with matched delays remain elusive. While transmission delays in brain networks change during development, the process by which delays are tuned for efficient transmission is unknown. Here, we demonstrate that the well-known Hebbian learning rule can provide a mechanism for selecting connections with delays that match the period of network oscillations. We consider a reciprocally connected bi-layer network of excitatory and inhibitory neurons that generate network-level oscillations spontaneously or in response to external stimuli. When exposed to spiketiming-dependent plasticity (STDP), the network self-organizes to potentiate connections with delays matching the oscillation period, while depressing those with non-matching delays. Our findings shed light on how transmission delays may evolve during learning and development to optimize the organization of brain networks for efficient signal transmission.