Fig. 4: Lifespan development of functional gradients with respect to canonical resting-state networks.
From: Functional hierarchy of the human neocortex across the lifespan
a, Population-level GAMM fits of mean gradient value versus age within each Schaefer 7-network parcel for the SA, VS and MR axes; maximum or minimum lifespan derivatives of mean network-wise gradient values are marked on the trajectories. b, GAMM fit of within-network dispersion versus age, computed as the mean Euclidean distance between each vertex’s coordinate in the SA–VS–MR embedding space and the corresponding network centroid. DorsAttn, dorsal attention; SalVentAttn, salience and ventral attention; SomMot, somatomotor. c, Network maturation ages for gradient value and dispersion. For each network and metric, maturation age is defined as the age at which the fitted trajectory reaches its extremum (maximum or minimum, as appropriate); points denote the posterior median extremum age and horizontal error bars denote 95% highest-density intervals (HDIs) estimated from coefficient resampling. d, GAMM-fitted gradient embedding plots coloured by the Schaefer 7-network parcellation at selected ages, shown in the SA–VS (top) and SA–MR (bottom) planes; in the SA–MR view, MR polarity is oriented left to right from representation (M) to modulation (R). In a,b, n = 3, 972 gradient sets; shaded ribbons indicate approximate 95% confidence intervals of the fitted mean.
