Adolescent brain maturation involves structural changes effecting a shift in excitation/inhibition (E/I) balance, yet the functional implications of these changes remain unclear. One implication is a shift with respect to criticality. Adult brains, at rest, operate near a critical phase transition, at the boundary between an active, excitation-dominant phase, and an absorbing, inhibition-dominant phase. Special properties emerge when neural systems are balanced at criticality, including maximal susceptibility to perturbation, dynamic range, and information transmission. Thus, a clear picture of how adolescent brain maturation affects E/I balance and criticality is needed to understand how maturational processes shape cognition. Here, we leverage the dynamical properties of longitudinally-collected resting-state EEG recordings during N = 310 sessions from 169 healthy human participants ranging in age from 10 to 33 years old to quantify E/I and proximity to criticality. We find that adult brains operate closer to criticality, including spectrally-widespread increases in long-range temporal correlations and amplitude bistability. We also find band-specific changes in excitation versus inhibition whereby the mechanisms driving low-frequency (theta to ) oscillations shift towards lower E/I — possibly because of increasing inhibition — while the mechanisms driving high-frequency (gamma) oscillations shift towards higher E/I, possibly because of decreasing inhibition. Opening eyes shifts brains towards lower E/I, and these state-dependent shifts are larger in adults. We simulate developmental effects with a neural mass model of coupled excitatory and inhibitory neurons providing a parsimonious account of how changes in brain dynamics could arise as a function of changes in local connectivity of excitatory and inhibitory neurons. Results indicate developmental movement towards criticality, and greater adaptability to state-specific demands, through adolescence to adulthood, reflecting changes in E/I balance with implications for cognitive development.
Depression subtype classification from social media posts: few-shot prompting vs. fine-tuning of large language models
BackgroundSocial media provides timely proxy signals of mental health, but reliable tweet-level classification of depression subtypes remains challenging due to short, noisy text, overlapping symptomatology,



