Figure 10

a) Comparison (through the average comparison measure \(\bar{B}\)) between the target cluster partition and the partition obtained from network simulations with connectivity matrix \(A_{21}\) and a white Gaussian noise term \(\eta _i(t)\) added to every node’s excitatory subpopulation input; the result is obtained for \(\sigma =10^{-3}\). \(\sigma _{\eta }\) is the noise signal standard deviation (see Eq. (7) in Methods). The gray dashed line marks the value of \(\bar{B}\) obtained by simulating the network with the original structural connectivity matrix \(A_0\) and without noise for the same level and same \(\sigma\) value (see Fig. 6). b) Comparison (through the average comparison measure \(\bar{B}\)) between the target cluster partition and the partition obtained from network simulations with connectivity matrix \(A_{21}\) and heterogeneous nodes. Heterogeneity is introduced by sampling the parameters \(w_{EE}\), \(w_{IE}\), \(w_{EI}\) (see Eq. (7) in Methods) from Gaussian distributions with means \(\mu _{w_{EE}}=3.5\), \(\mu _{w_{IE}}=2.5\) and \(\mu _{w_{EI}}=3.75\) and standard deviation \(\sigma _w\). The result is obtained for \(\sigma =10^{-3}\). The gray dashed line marks the value of \(\bar{B}\) obtained by simulating the network with the original structural connectivity matrix \(A_0\) and homogeneous nodes for the same level and same \(\sigma\) value.