Reconfiguration of functional brain networks and metabolic cost converge during task performance

Abstract

The ability to solve cognitive tasks depends upon adaptive changes in the organization of whole-brain functional networks. However, the link between task-induced network reconfigurations and their underlying energy demands is poorly understood. We address this by multimodal network analyses integrating functional and molecular neuroimaging acquired concurrently during a complex cognitive task. Task engagement elicited a marked increase in the association between glucose consumption and functional brain network reorganization. This convergence between metabolic and neural processes was specific to feedforward connections linking the visual and dorsal attention networks, in accordance with task requirements of visuo-spatial reasoning. Further increases in cognitive load above initial task engagement did not affect the relationship between metabolism and network reorganization but only modulated existing interactions. Our findings show how the upregulation of key computational mechanisms to support cognitive performance unveils the complex, interdependent changes in neural metabolism and neuro-vascular responses.

No metabolic “default mode” of human brain function

“…Importantly, both modalities also demonstrated an increase in activity with an increase in task demand (manipulation > maintenance) across the task-positive networks ( Fig. 2a, bottom row), in line with recent results from a visuo-spatial task 44 . Across contrasts, this shows that changes in glucose metabolism parallel BOLD signal increases in task-positive regions, providing important initial support for the feasibility of the method and current design.…”

“…This functional PET (fPET) method relies on a modification of the traditional bolus injection paradigm into a slow infusion protocol that delivers a constant plasma supply of FDG to the cell, such that a dynamic change in the slope of the time activity curve is proportional to the rate of CMRglu. Using this method, transient metabolic changes over task blocks that are only a few minutes long have been detected in task-relevant regions during visual stimulation, hand movement as well as visuo-spatial reasoning [39][40][41][42][43][44] , demonstrating marked spatial overlap with BOLD responses during the same task 44 .…”

Stress-sensitive inference of task controllability

“…By defining the most appropriate state space dynamically, controllability estimation improves predictions about the future states of one's environment and can therefore contribute to maximize utility when reward or punishment rates depend on such predictions. By promoting reliance on a simpler spectator model when the environment is deemed uncontrollable, it can also minimize the metabolic cost and subjective effort associated with controlled action selection (Hahn et al, 2020;Westbrook et al, 2020). These hypotheses could be tested directly by introducing reinforcers in the explore-andpredict task, but it is already worth to note that the controllability-dependent arbitration logics can readily explain why Pavlovian (equivalent to SS') and instrumental (equivalent to SAS') learning mechanisms are respectively promoted uncontrollable and controllable contexts (Dorfman and Gershman, 2019).…”

Simultaneous BOLD-fMRI and constant infusion FDG-PET data of the resting human brain

“…FDG-PET has a spatial resolution of around 4 mm 12 and, until recently, a temporal resolution that was effectively equal to the scan duration – around 10–40 minutes. However, recent developments in radiotracer delivery 13 , 14 have resulted in substantial improvements in FDG-PET temporal resolution, reducing it down to 60 sec 14 – 18 or less [12 sec 19 ; 16 sec 13 , 20 ; 30 sec 21 ]. This method, known as ‘functional’ PET (fPET) administers the radiotracer throughout the scan, either as a constant infusion 14 or hybrid bolus/infusion 19 .…”

“…Examples of important validation work yet to be performed includes test-retest reliability (which can be difficult in humans for ethical reasons due to biosafety constraints related to radioactivity exposure), and whether the FDG in a continuous infusion protocol acts as a radiotracer or is more analogous to a radioligand 28 . Examples of re-use of the Monash rsfPET-MR dataset may include development of data analysis techniques 18 , synergistic data fusion techniques 29 , discoveries about the relationship between glucose uptake and the haemodynamic response 17 , as well as revelations about the fundamental basis of energy use in the human brain 21 , 30 – 32 .

Fig.

…”