Fig. 6: Impact of critical state on structural and functional connectivities.
From: A hierarchy of time constants and reliable signal propagation in the marmoset cerebral cortex
a Depiction of structural connectivity (SC) and functional connectivity (FC) under various scenarios. Top panel: Plot representing structural connectivity (SC) as per the FLN matrix. Middle Panel: Visualization of FC, measured by the correlation coefficient between resting-state neural activity from different areas, with the presence of composite gradient. Bottom Panel: Visualization of FC in the absence of composite gradient. b Dissimilarity between FC and SC as a function of gradient slope. Top panel: Illustration of two distinct approaches to modifying the gradient slope, depicted through a color gradient from dark to light, indicating a change in slope from 1 to 0. Left, the adjustment of gradient slope with varying excitability. Right, the adjustment of gradient slope with constant average excitability. Bottom panel: The dissimilarity increases with the slope of gradient, with or without average excitability held constant. Here the dissimilarity is quantified as 1 − ∣ρ∣, where ρ is the Pearson correlation between FC and SC. c Heatmap of FC-SC dissimilarity across different brain areas, shown in marmoset brain parcellation. Left: FC-SC dissimilarity produced by the control model with composite gradient. Right: the model result with composite gradient removed. d Left, correlation between the FC-SC dissimilarity for each brain area ranked by its composite gradient of excitation in an ascending order (Pearson r = 0.41, p = 2.12 × 10−3). Right, absence of this correlation when the gradient is removed (r = 0.09, p = 0.491). e FC-SC dissimilarity across seven functional subnetworks of the marmoset brain. Left: the control model result. Right: The model result with gradient removed. Each dot represents the FC-SC dissimilarity of an individual brain area. Larger overlaid points show the mean for each subnetwork, and error bars denote ±1 standard deviation around the mean. Sample sizes (n, number of distinct brain areas per subnetwork) from bottom to up are n = 7, 11, 6, 6, 2, 14, and 4.
