Structural and information-theoretic complexity measures of brain networks: Evolutionary aspects and implications

Abstract

The evolutionary lineage of neuronal phenotype is notably complex even within a limited number of species. One of the approaches resides in the realm of complex network theory. The theory reduces the connectomic data into a hallmarked set of few parameters, some of which might be correlated with a suitably chosen phylogenetic marker. In this first-of-its-kind attempt, interspecific variations of two structural complexity measures (i.e., clustering coefficient and centrality) along with two independent information-theoretic measures (i.e., von Neumann entropy and multifractality) are investigated to decipher any hidden evolutionary signature considering four mammalian connectomes (i.e., felis catus, mus musculus, macaca mulatta, and homo sapiens). All network complexity measures partially corroborate with the phylogenetic order. Nevertheless, monotonicity of the measures with the chosen phylogenetic marker of genome size has been majorly violated because of the mus musculus data point. On the other hand, von Neumann entropy was found to exhibit an allometric scaling behavior with the community structure of all connectomes (p<0.0001, and R2>0.95). The respective scaling exponent was noted to be monotonic with the genome size. Singularities of the real connectomes were also investigated upon carrying out a similar analysis in three equivalent synthetic network models.