Dynamical Activity Patterns of High-frequency Oscillations and Their Functional Roles in Neural Circuits

Abstract

Oscillations are widely observed in brain activities, but their spatiotemporal organisation properties and functional roles remain elusive. In this thesis, we first investigate gamma oscillations (30 - 80 Hz) by combined empirical and modelling studies. Array recordings of local field potentials in visual motion-processing cortical area MT of marmoset monkeys reveal that gamma bursts form localised patterns with superdiffusive propagation dynamics. To understand how these gamma burst patterns emerge, we investigate a spatially extended, biophysically realistic circuit model that quantitatively captures the superdiffusive propagation of gamma burst patterns as found in the recordings. We further demonstrate that such gamma burst dynamics arise from the intrinsic non-equilibrium nature of transitions between asynchronous and propagating wave states. These results thus reveal novel a spatiotemporal organisation property of gamma bursts and their underlying mechanisms. We further illustrate that such non-equilibrium transitions in the spatially extended circuit model mechanistically account for a range of dynamical properties of sharp-wave ripples (100-250 Hz) and that sharp-wave ripples with superdiffusive dynamics underlie efficient memory retrievals. Furthermore, by incorporating short-term synaptic plasticity into the spatially extended circuit model, we find that in the dynamical regime near the transition of circuit states, large endogenous fluctuations emerging from the circuit can quantitively reproduce and explain the variability of working memory across items and trials as found experimental studies. In addition, our circuit model reproduces and explains several other key neural and behavioural features of working memory, such as the capacity limit of working memory and gamma bursts that are coupled to theta oscillations. These results establish a dynamical circuit mechanism of working memory and provide novel experimentally testable predictions.