The control costs of human brain dynamics

Abstract

The human brain is a complex system with high metabolic demands and extensive connectivity that requires control to balance energy consumption and functional efficiency over time. How this control is manifested on a whole-brain scale is largely unexplored, particularly what the associated costs are. Using network control theory, here we introduce a novel concept, time-averaged control energy (TCE), to quantify the cost of controlling human brain dynamics at rest, as measured from functional and diffusion MRI. Importantly, TCE spatially correlates with oxygen metabolism measures from positron emission tomography, providing insight into the bioenergetic footing of resting state control. Examining the temporal dimension of control costs, we find that brain state transitions along a hierarchical axis from sensory to association areas are more efficient in terms of control costs and more frequent within hierarchical groups than between. This inverse correlation between temporal control costs and state visits suggests a mechanism for maintaining functional diversity while minimizing energy expenditure. By unpacking the temporal dimension of control costs, we contribute to the neuroscientific understanding of how the brain governs its functionality while managing energy expenses.

Modeling the neurocognitive dynamics of language across the lifespan

“…Indeed, compared to older adults, we found that younger adults capitalize on the cooperation or synergy between the DMN and FPN to enhance cognitive efficiency (Luppi et al, 2024;Xia et al, 2022). This is consistent with studies suggesting that the communication between higher-order networks on the sensorimotorassociative hierarchy, such as the DMN and FPN (Margulies et al, 2016), is more efficient at handling the high metabolic demands associated with long-range or rich club connectivity (Ceballos et al, 2024;Roy et al, 2017). Relatedly, our results are also consistent with the DECHA model (Spreng & Turner, 2019), proposing that reduced DMN suppression may result in a more inflexible modulation of the connectivity between the DMN and FPN in response to task challenges, a fortiori mediating cognitive decline in older adults, especially in LP.…”

“…dence showing reduced functional connectivity of long-range connections in older adults(Sala-Llonch et al, 2014) and may offer a metabolic explanation for the onset of cognitive decline observed in midlife in most cognitively demanding tasks (e.g., seeCeballos et al, 2024 for a study on the costs of brain dynamics).…”

Modeling the Neurocognitive Dynamics of Language across the Lifespan

“…Indeed, compared to older adults, we found that younger adults capitalize on the cooperation or synergy between the DMN and FPN to enhance cognitive efficiency (Luppi et al, 2024; Xia et al, 2022). This is consistent with studies suggesting that the communication between higher-order networks on the sensorimotor-associative hierarchy, such as the DMN and FPN (Margulies et al, 2016), is more efficient at handling the high metabolic demands associated with long-range or rich club connectivity (Ceballos et al, 2024; Roy et al, 2017). Relatedly, our results are also consistent with the DECHA model (Spreng & Turner, 2019), proposing that reduced DMN suppression may result in a more inflexible modulation of the connectivity between the DMN and FPN in response to task challenges, a fortiori mediating cognitive decline in older adults, especially in lexical production.…”

“…In comparison, older adults seem to adopt a more “energy-efficient” approach, substituting the reallocation of long-range connections between subsystems with short-range connections within subsystems, thus lowering the metabolic demands needed to achieve compensatory integration. This joins previous evidence showing reduced functional connectivity of long-range connections in older adults (Sala-Llonch et al, 2014) and may offer a metabolic explanation for the onset of cognitive decline observed in midlife in most cognitively demanding tasks (e.g., see Ceballos et al, 2024 for a study on the costs of brain dynamics).…”