Fig 5. Cell assemblies and connectivity.

a) Cross-correlation matrix C(ν_i, ν_j) of the firing rates organized according to the clusters generated via the k-means algorithm with k = 15, the clusters are ordered as in Fig 1(c) from the highest to the lowest correlated one. b) Connectivity matrix ${\mathcal{C}}_{ij}$ with the indices ordered as in panel a). Here, a black (copper) dot denotes a 1 (0) in ${\mathcal{C}}_{ij}$, i.e. the presence(absence) of a synaptic connection from j to i. c) Average cross-correlation 〈C〉_ml among the elements of the matrix block (m, l), versus the probability p_ml to have synaptic connections from neurons in the cluster l to neurons in the cluster m. d) 〈C〉_ml versus the block averaged similarity metrics e_ml. Black squares indicate the blocks along the diagonal delimited by black borders in panel a) and b); blue triangles denote the ten blocks with the lowest 〈C〉_ml values, which are also delimited by blue edges in a) and b). The vertical red dashed line in panel c) denotes the average probability to have a connection p = 5% and in panel d) the value of the metrics e_ml averaged over all the blocks. The black solid line in panel c) is the linear regression to the data (〈C〉_ml ≈ 0.15–3.02p_ml, correlation coefficient R = −0.72). Other parameters as in Fig 1.