Fig. 3: Putative global dynamical properties of the fly central nervous system.

a, Magnitude of the top 1,000 eigenvalues of the putative effectome (scaled matrix of signed synaptic counts extracted from the connectome) and power-law fit. b, Eigenvalues plotted in the complex plane. c, Correlation of eigenvectors sorted by associated eigenvalue magnitude. d, Median (black), 99th percentile (dark grey) and maximum (light grey) correlation between each eigenvector and all other top 1,000 eigenvectors, showing that the first 10–20 eigenvectors are nearly orthogonal to other eigenvectors and the correlation between other eigenvectors is highly sparse. e, Per neuron eigenvector loadings for the first (top) and 50th (bottom) eigenvector. f, Concentration of eigenvector loadings, quantified by the number of neurons needed to account for 75% of the power of the eigenvector. g, Anatomical renderings of neurons needed to account for 75% of the eigenvector’s loading power. Eigenvectors are drawn from the top 100 eigenvectors, and each column displays eigenvectors with neurons predominantly associated with either visual, olfactory or motor/navigation anatomical locations.