What lies beneath the brain’s networks?

Why do some brain regions tend to function together? A new study explores this question by linking the organization of functional networks to the cellular and molecular architecture of the brain.

When we look at a functional MRI scan, the brain appears as a map of networks. Certain regions show coordinated fluctuations in activity and therefore form distinct functional systems. The default mode network, sensorimotor networks, visual networks and cognitive control networks each reflect a particular organization of brain activity. But one question remains: what gives these networks their organization? Why do some regions tend to function together?

This is the question researchers explored in a study published in 2025 in Nature Communications. By comparing the organization of brain networks with different biological characteristics of the brain, they sought to understand what lies beneath the patterns of activity detected by MRI.

Beyond the maps of brain activity

The researchers examined 53 reproducible intrinsic connectivity networks identified from functional MRI data. These networks covered several major functional domains, including sensorimotor, visual, auditory and cognitive control systems, the default mode network, as well as subcortical and cerebellar networks. The networks were derived from the NeuroMark functional MRI template, based on data from 1,828 young adults in the Genomics Superstruct Project and Human Connectome Project.

They then compared this organization with the distribution of different cell types in the brain. Some neuronal populations showed specific spatial correspondences with particular functional networks. It is important, however, to be precise about what an “association” means here. The researchers did not show that a particular cell type determines the function of a network on its own. Rather, they found that the distribution of certain cell populations partly corresponds to the spatial organization of specific functional networks.

The researchers also examined the brain at a finer molecular scale by looking at gene expression. Using data from resources including the Allen Human Brain Atlas, together with single nucleus RNA sequencing data, they examined how molecular and cellular characteristics were distributed across different brain regions. This type of approach, often described as imaging transcriptomics when transcriptomic information is linked to neuroimaging, makes it possible to compare maps of gene expression with maps obtained through brain imaging.

The results indicate that functional networks are associated with distinct patterns of molecular organization rather than being distributed randomly. In particular, genes and molecular profiles related to neuronal and synaptic functions contributed to differences observed between networks. The researchers therefore identified molecular signatures that varied across the brain’s major functional organizations.

These findings should not be interpreted as meaning that a particular gene “controls” a brain network. Higher expression of certain genes in a region does not mean that those genes directly create the network detected by MRI. The study identifies statistical correspondences between brain maps generated using different methods. The significance of these findings lies instead in showing that functional differences between networks are accompanied by differences in their molecular environment. Gene expression therefore provides another level of information for understanding why some brain regions have different biological and functional properties.

The brain’s chemical signature

After examining cells and genes, the researchers turned to another fundamental aspect of neuronal communication: the chemical systems through which neurons interact. This involves not only neurotransmitters but also the receptors and transporters that allow cells to respond to and regulate these chemical signals.

To investigate this dimension, the researchers used 19 maps representing receptors and transporters from nine neurotransmitter systems. These maps were derived from PET imaging data collected from more than 1,200 healthy individuals. This allowed the researchers to compare the distribution of these neurotransmitter systems with the organization of different brain networks.

The results showed that some networks had neurochemical profiles that were more strongly associated with them than others. Among the 53 networks examined, 16 showed statistically significant associations with the receptor and transporter maps. The relative contribution of different neurotransmitter systems therefore varied across networks.

These findings also illustrate why it is too simplistic to speak of “the effect” of a neurotransmitter in isolation. Dopamine, serotonin, glutamate and GABA do not function like simple switches. Their effects depend, among other factors, on the receptors they act on, the cells expressing those receptors and the circuits in which they operate.

The researchers also examined another essential dimension of brain function: energy metabolism. Neurons require a continuous supply of energy to maintain their activity and support communication between cells. Mitochondria, which play a central role in cellular energy production, were therefore included in the analysis. Some mitochondrial characteristics showed associations with functional networks, particularly within subcortical systems. These findings were more limited than those observed for cellular characteristics and neurotransmitter systems.

Cells, genes, chemistry and energy therefore begin to point toward a common picture: the networks observed with MRI are embedded in a biological system far more complex than a simple map of connections would suggest.

🔗 Explore further: What dreams actually reveal

When biology meets brain networks

A central question remained: are these biological differences related to the way brain networks communicate with one another?

The researchers compared the biological profiles of the networks with their functional connectivity. They found that networks with more similar cellular and molecular characteristics also tended to show stronger functional relationships. This correspondence was particularly apparent when several biological dimensions were considered together.

The same relationship was observed when the researchers examined dynamic functional connectivity, referring to patterns of communication between networks that can change over time. The study identified several dynamic connectivity states whose organization was associated with cellular and molecular similarity networks.

These findings add another layer to the concept of a brain network. Networks are distinguished not only by their patterns of activity or by the regions they contain. They also show characteristic biological profiles. This does not mean that these biological characteristics mechanically determine connectivity. The study demonstrates statistical relationships rather than a simple causal chain. The biological architecture may help constrain the conditions under which particular functional organizations are more likely to emerge, without determining them on its own.

The researchers then sought to connect these different levels of organization to cognition. They used 123 probabilistic cognitive and behavioral maps derived from meta analytic data available through Neurosynth. Some of the analyses suggested that functional connectivity may act as an intermediary between the biological characteristics of brain regions and their cognitive organization.

The different scales of brain organization therefore do not appear to operate independently. Cellular and molecular characteristics, network organization and cognitive functions show relationships that can be examined within a common framework. This is one of the major strengths of the approach. Rather than studying genes, cells, circuits, networks and cognition separately, it seeks to understand how these different levels are linked within the same system.

🔗 Discover more: Timeless regeneration: Uncovering the brain’s stem cell secrets

Beyond connectivity

The main contribution of this study is therefore conceptual. It encourages us to move beyond the idea of the brain as a simple map of connections. Behind the networks observed with MRI are neurons, synapses, receptors, glial cells, energy systems and a complex molecular architecture.

Understanding a brain network therefore requires more than asking which regions communicate with one another. It also requires understanding the biological properties that make this communication possible.

This does not mean that genes directly determine our thoughts, or that a particular neurotransmitter corresponds to a specific mental function. The brain does not operate through such a linear chain of causation. It is better understood as a system organized across multiple levels, from genes to cells, from cells to circuits, from circuits to networks, and from networks to cognitive functions and behavior.

The study by Feng and colleagues does not provide a definitive explanation of how this organization arises. Instead, it adds another piece to a central question in neuroscience: how can we connect what happens at the microscopic level with what we observe across the brain as a whole?

MRI maps reveal patterns of brain activity, but they capture only part of the picture. Behind every network lies living tissue, with its cells, chemistry, energy demands and complex biological organization. Understanding brain function ultimately means bringing these different levels together.