Basis of executive functions in fine-grained architecture of cortical and subcortical human brain networks

Abstract

Theoretical models suggest executive functions are supported by both domain-general and domain-specific processes. While some brain imaging studies claim executive tasks recruit a domain-general multiple-demand (MD) brain system, many studies argue the spatially coarse results of traditional imaging methods have blurred fine-grained functionally fractionated domain-specific systems into one. To address this challenge, we scanned participants using the high spatial resolution multimodal MRI approach of the Human Connectome Project while they performed tasks targeting executive demands of updating, shifting and inhibition. The results show that different executive activations overlap, at the single subject level, within MD regions. Critically, each task's topography shifts within MD regions to form a unique intersection with adjacent fine-grained resting-state networks. In this intersection, the strongest activations arise at neurobiologically defined network borders. Outside cerebral cortex, matching results are seen in circumscribed regions of the caudate nucleus, thalamus and cerebellum. Using precise imaging methods, these results suggest a novel framework whereby partially-specialised networks recruit neighbouring MD areas to generate distinct executive functions.

A hierarchical atlas of the human cerebellum for functional precision mapping

“…Consistent with the characteristics of the cortical multi-demand system [27], all regions showed significant activation during executive tasks (n-back, switch and stop tasks), and increased activity especially with high difficulty. Nonetheless, there was some functional specialization across the regions.…”

“…The first four datasets each comprised a broad battery of tasks tapping into cognitive, motor, perceptual, and social functions: (1) The Multi-Domain Task Battery dataset (MDTB, [3]), (2) a high-resolution version of the MDTB (High-res MDTB; not yet published), (3) the Nakai & Nishimoto dataset [44], and the (4) The Individual Brain Charting (IBC) dataset [45,46]. We also included three further datasets to obtain a better description of the domains of motor movements and executive functions: (5) the working memory (WM) dataset [43] which included finger movements and a forward / backwards digit span task; (6) the Multi-Demand dataset [27] which included a no-go, N-back, and task-switch task ; and (7) the Somatotopic dataset [25] which probed foot, hand, glutes, and tongue movements. Finally, we used the resting-state fMRI dataset Unrelated 100 subjects, which is made publicly available in the Human Connectome Project (HCP) S1200 release [47].…”

A hierarchical Bayesian brain parcellation framework for fusion of functional imaging datasets

“…Because there is no single, large task-based dataset that would cover all of the mental functions, we developed here a framework that allows us to fuse data from a growing number of deep-phenotyping task-based datasets with fewer participants (King et al, 2019, Pinho et al, 2018, Nakai and Nishimoto, 2020, Assem et al, 2022. To make data fusion feasible in a Bayesian framework, we deployed a series of emission models, each one learns the specific characteristics of the corresponding dataset, including the expected response for each brain region and their variability.…”

High gamma activity distinguishes frontal cognitive control regions from adjacent cortical networks

“…We recorded LFP signals using ECoG from the lateral frontal surface of human patients undergoing awake craniotomies for tumor resection. We used a canonical demand on executive control, a contrast of cognitive switching versus simple counting, similar to many manipulations known to recruit FPN regions in fMRI studies ( Assem et al., 2020 , 2022 ; Fedorenko et al., 2013 ; Shashidhara et al., 2019 , 2020 ). The results revealed a circumscribed frontal region that shows increases in HG power.…”