Information Spillover in Resting Memory and Working Memory

Abstract

Our fMRI study investigates the dependencies between brain areas using directed spillover indices estimated from vector autoregressive models that recognize dynamics in the network. Spillover analysis examines how brief neural signal bursts evolve over time and how these dynamic changes can predict past and future brain states across different regions of the brain network. Information spillover was assessed using a test-retest design to estimate the neural memory capacity across distinct cognitive states. A dorsolateral prefrontal centered system (DLPFC) demonstrates neural memory capacity at rest, labeled "resting memory". Resting memory contains roughly 9 times more neurocognitive dependencies (spillover) as the difference in spillover between working and resting brains, suggesting that resting brains exhibit substantial neural memory capacity. The transitioning from "resting memory" to "working memory" is initiated by a right inferior fontal (IFG) centered system which connects to the DLPFC centered system when relevant information is detected in the outside world and also inhibits self-referential feedback in parietal cortices. Spillover between the IFG and DLPFC centered systems facilitate a smooth transition in attention from events that take place outside the brain to (sustained) representations of external events within the brain.