"a 379-nodes large-scale BNMs were constructed from person-specific white matter connectomes estimated with dwMRI tractography. In addition, a simplified network with only two nodes (but identical node dynamics) was used to create E/I-ratio tuning curves (Fig. 4). b In previous BNM studies long-range white matter coupling from excitatory to inhibitory populations was often absent. Adding these connections allowed to tune the relative strength of long-range excitatory-to-excitatory versus long-range excitatory-to-inhibitory connections, enabling to precisely tune the E/I-ratio of synaptic inputs between each pair of BNM nodes. Importantly, setting the E/I-ratio allowed to monotonically and smoothly control the FC between all nodes (Fig. 3a). Underlying predicted fMRI time series, the E/I-ratio allowed to smoothly tune synchronization and amplitude of synaptic currents (Fig. 4). c By systematically tuning E/I-ratios, the fit between simulated and empirical FC can be increased until full similarity (Fig. 3b, c). d Upon fitting each participant’s BNM with their empirical FC, each BNM was coupled with a smaller scale frontoparietal circuit for simulating DM and WM. Subpopulations in prefrontal cortex (PFC) and posterior parietal cortex (PPC) are mutually and recurrently coupled to encode two decision options A and B. For example, evidence for option A recurrently excited the populations APPC and APFC (red connections) while it led to an inhibition of the populations BPPC and BPFC (blue connections). Importantly, instead of independent noise, we used the activity of the PFC and PPC regions of the 379-nodes large-scale network to drive the DM circuit, which allowed to analyze how local decision-making and working memory performance can be modulated by large-scale brain network topology. Panel a is adapted from ref. 77. and used under a CC BY 4.0 license (https://creativecommons.org/licenses/by/4.0/)." Study featured in '2023 in science' here.