Fig. 6: Multiscale modeling: coupling PFC and PPC nodes of the person-specific BNMs with the corresponding modules of the generic DM circuit. The models of subjects with higher PMAT24_A_CR (fluid intelligence) made fewer mistakes, but were slower, echoing the empirically observed trade-off.
From: Learning how network structure shapes decision-making for bio-inspired computing
a Distribution of significant correlations between mean input of all BNM nodes and PMAT24_A_CR (p < 0.05 for 35 of 379 nodes), respectively PMAT24_A_RTCR (p < 0.05 for 26 of 379 nodes) over all N = 650 models. b, c Group-average PMAT24_A_CR versus DM performance (r = 0.77, \(p=7.2\times {10}^{-5}\)), respectively DM time (r = 0.69, \(p=7.2\times {10}^{-4}\)), for an exemplary combination of PFC and PPC nodes. Data are presented as mean values +/− SD over all N = 650 models each simulated 100 times with different random number generator seeds. d Distribution of significant correlations between group-average PMAT24_A_CR and DM time (p < 0.05 for 57 of 90 possible combinations), respectively, DM performance (p < 0.05 for 19 of 90 possible combinations) over all N = 650 models. Including only correlations that remained significant after controlling for multiple comparisons using the Benjamini-Hochberg procedure with a False Discovery Rate of 0.1.
