Connectome of a model local cortical circuit flexibly shapes layer-dependent multi-frequency oscillations

Abstract

The role played by interlayer connections in shaping local responses and their long-range coupling has not yet been fully elucidated. Here, we analyze a rate model of a canonic local circuit with realistic anatomy. We find that this circuit generates a rich repertoire of possible dynamical states, including an oscillatory regime in which gamma-and beta-oscillations dominate in superficial and deep layers, respectively, in agreement with experimental observations. This regime stems from non-linear inter-layer interactions, independently from intrinsic resonance properties of distinct layers. Moreover, by connecting two local circuits via cortico-cortical projections, the emergent phase differences define a flexible and frequency-dependent inter-areal hierarchy. Such dynamic patterns generally do not arise in randomized circuits, and the compatible connectomes are rare, although not unique. Altogether, these results suggest that inter-layer connectivity is homeostatically regulated to make local circuits fit to integrate and multiplex signals from several sources in multiple frequency bands.\n\nAuthor SummaryThe local circuit of mammalian cortex presents a characteristic multilayered structure, with feedforward (and feedback) cortico-cortical connections originating from (and targeting) distinct and well defined layers. Here, we model how such a structure fundamentally shapes the dynamical repertoire of local cortical oscillatory states and their long-range interaction. Experimental evidence, matched by our simulations, suggests that different cortical layers oscillate at different frequencies and that neuronal oscillations at different frequencies are exploited for communication in different directions. While this laminar specificity of oscillations is often explained in terms of multiple inhibitory populations with different resonance properties, we show here that it could alternatively emerge as a byproduct of the collective local circuit dynamics. Our modelling study indicates furthermore that the empirically observed multi-frequency oscillatory patterns cannot be reproduced in presence of an arbitrary interlayer connectivity. In this sense, therefore, we believe that the adopted connectome, derived from neuroanatomical reconstructions, is \"special\". Nevertheless, it is not unique, since other, very different connectomes may also lead to a matching dynamical repertoire. This suggests that a multiplicity of non-random canonical circuit templates may share largely overlapping functions, robustly achieved and maintained via functional homeostasis mechanisms.