Metabolic brain networks switch between a sparsely connected baseline and highly integrated states to support cognition

Abstract

The human brain achieves cognitive flexibility by rapidly switching between large-scale functional network states. While state switching is assumed to be energetically demanding, the direct metabolic foundation underpinning state switching has not been characterised. Here, we combine functional FDG-PET (fPET) imaging and sliding-window analyses to characterise metabolic network state switching in 85 adults (20-89 years). Four recurring metabolic network states were identified: a highly prevalent, sparsely connected baseline alongside three transient, globally integrated network states of cognitive control and attention. Cognitive performance relied on the capacity to mobilise the integrated associative system states. Flexible network switching was enabled by the moment-to-moment dynamic range of the underlying regional glucose signals. Neurotransmitter analyses revealed a low-dimensional neurochemical hierarchy that mapped to the dynamic network states, anchored by a dominant axis of endocannabinoid, metabolic, serotonergic and GABAergic systems. The dynamic metabolic network architecture was also attenuated in older adults, who exhibited a loss of network flexibility and were anchored to the sparsely connected baseline state. These findings reveal that functional network state switching is an emergent property of the brains metabolic architecture. This understanding may provide critical new insights into the metabolic basis of brain ageing, neurodegeneration and psychiatric conditions.