Learning to be on time: temporal coordination of neural dynamics by activity-dependent myelination

Abstract

Activity-dependent myelination is the mechanism by which myelin changes as a function of neural activity, and plays a fundamental role in brain plasticity. Mediated by structural changes in glia, activity-dependent myelination regulates axonal conduction velocity. It remains unclear how neural activity impacts myelination to orchestrate the timing of neural signaling. We developed a model of spiking neurons enhanced with neuron-glia feedback. Inspired by experimental data and use-dependent synaptic plasticity, we introduced a learning rule, called the Activity-Dependent Myelination (ADM) rule, by which conduction velocity scales with firing rates. We found that the ADM rule implements a homeostatic control mechanism that promotes and preserves synchronization. ADM-mediated plasticity was found to optimize synchrony by compensating for variability in axonal lengths by scaling conduction velocity in an axon-specific way. This property was maintained even when the network structure is altered. We further explored how external stimuli interact with the ADM rule to trigger bidirectional and reversible changes in conduction delays. These results highlight the role played by activity-dependent myelination in synchronous neural communication and brain plasticity.