RT Journal Article
SR Electronic
T1 Mechanisms of distributed working memory in a large-scale model of the macaque neocortex
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 760231
DO 10.1101/760231
A1 Jorge F. Mejias
A1 Xiao-Jing Wang
YR 2019
UL http://biorxiv.org/content/early/2019/09/08/760231.abstract
AB Working memory, the brain’s ability to retain and manipulate information internally, has been traditionally associated with persistent neural firing in localized brain areas such as those in the frontal cortex (Fuster 1973, Funahashi et al., 1989; Goldman-Rakic 1995; Romo et al., 1999; Rigotti et al., 2013; Kopec et al., 2015; Inagaki et al., 2019). However, self-sustained neural persistent activity during working memory is present in multiple brain regions (Romo et al., 2004; Christophel et al., 2017; Leavitt et al., 2017; Sreenivasan and D’Esposito, 2019), the underlying mechanism is unknown. We developed an anatomically constrained large-scale computational model of the macaque cortex endowed with a macroscopic gradient of synaptic excitation, to investigate the origin of distributed working memory representation. We found that long-range inter-areal reverberation can support the emergence of persistent activity patterns across multiple cortical regions, by virtue of a robust bifurcation in space, even when none of isolated local areas is capable of generating persistent activity. The model uncovered a host of distinct persistent activity patterns (attractor states), and provides experimentally testable predictions that cannot be explained in local circuit models. Simulating cortical lesions reveals that distributed activity patters are resilient against simultaneous lesions of multiple cortical areas, but depend on areas that form the core of the entire cortex. This work provides a theoretical framework for identifying large-scale brain mechanisms and computational principles of distributed cognitive processes.