How does the human brain’s electrical activity grow from childhood, peak in adulthood, and decline in older age? A multinational team has tackled this question by linking the brain’s “wiring diagram” and signal‑conduction speed to two familiar features of an electroencephalogram (EEG): the broadband background activity (ξ, pronounced “xi”) and the more rhythmic alpha waves. Their work, published in National Science Review , introduces a new model called Xi–αNET (“Xi–AlphaNET”) that explains how anatomical connections and nerve‑signal delays give rise to these patterns and how they change over the lifespan.
At the heart of the study is the HarMNqEEG dataset, a unique collection of resting‑state EEG recordings from 1,965 people aged five to 100 years. Participants were scanned in nine countries using 12 different EEG systems, and the data were harmonized to allow meaningful comparisons. Such breadth allowed the researchers to probe how the brain’s rhythms develop across an entire century of life.
Traditional analyses treat alpha waves and the background ξ signal as statistical patterns divorced from brain structure. Xi–αNET instead treats the aperiodic background (ξ) and the α‑rhythm as independent processes generated by the brain’s network. The model uses a myelination map derived from MRI to create a hierarchy of brain regions, then estimates how signals flow through this hierarchy. It shows that across the lifespan the broadband activity is localized in frontal regions and dominated by feedforward connections (from sensory areas upward), while the α‑rhythm is strongest in posterior sensory and sensorimotor regions and dominated by feedback connections (top‑down influences). This distinction echoes previous theories linking slower rhythms to long‑range feedback and faster rhythms to feedforward processing.
Xi–αNET also incorporates information about how long it takes for activity in one cortical region to reach another. These conduction delays are not measured directly by EEG; rather, they come from intracranial cortico‑cortical evoked responses, which provide priors on the time it takes for signals to travel between regions. The model then estimates a subject‑specific overall delay to align these prior delays to each individual. When the team examined how these delays vary with age, they found a U‑shaped trajectory—shorter delays in youth, stable midlife values, and longer delays in older age. Comparing this trajectory with independent MRI‑derived maps of myelination revealed that the curves closely match. In other words, the degree of insulation around nerve fibers (myelin) appears to set the pace of brain rhythms: faster conduction, reflecting heavier myelination, corresponds to higher alpha frequencies. The strong inverse relationship—peak alpha frequency declines as conduction delays lengthen—suggests that slowing alpha waves may be an accessible marker of declining white‑matter integrity in aging or disease.
Beyond its scientific insights, the work demonstrates the power of generative models—mathematical frameworks that explicitly link structure to function. The authors show that Xi–αNET produces reliable estimates of cortical activity, effective connectivity and subject‑specific conduction delays from routine EEG recordings. Such tools could pave the way for normative reference charts, against which individual deviations might flag developmental disorders, neurodegenerative diseases, or the effects of interventions. Preliminary analyses in the paper show that the model can detect the slowing of alpha rhythms in Parkinson’s disease, hinting at future clinical applications.
This study paints a new picture of brain rhythms: they are not free‑floating oscillations but reflections of the brain’s physical wiring and the efficiency of its signal highways. As lead author Ronaldo Garcia Reyes puts it, “By weaving together structural connections, conduction speed and electrical rhythms, we can start to understand how the brain’s architecture shapes its dynamics and why these dynamics change with age.”
National Science Review
Computational simulation/modeling