Theory of spontaneous persistent activity and inactivity in vivo reveals differential cortico-entorhinal functional connectivity

Abstract

Persistent activity is thought to mediate working memory. While such stimulus evoked persistence is well studied, mechanisms of internally generated or spontaneous persistence in vivo are unknown. Further, current theories based on attractor dynamics focus on elevated activity as a memory substrate, while little attention has focused on the role of inactivity attractors. Here, we present a mean field model of functional interaction between large cortical networks that predicts both spontaneous persistent activity (SPA) and inactivity (SPI); the latter has never been seen before in experiments or models. We confirm these predictions using simultaneously recorded neocortical local field potential (LFP) and the membrane potential (Vm) of identified excitatory neurons from several brain areas in vivo during slow oscillations, especially from layer 3 of the medial (MECIII) and lateral entorhinal cortex (LECIII), which show SPA and SPI. By matching model and experimental statistics, we predict the relative strength of internal and external excitation in the LECIII and MECIII networks. Our predictions match anatomical data. Further, the model predicts, and the experiments confirm, that SPA and SPI are quantized by cortical UDS and follow the statistics of a history dependent Bernoulli process. These convergent, theory-experiment results thus reveal the differential nature of cortico-entorhinal functional connectivity, resulting in a unique pattern of persistent activity and persistent inactivity, a novel and energetically efficient memory substrate.