Multivariate analysis of behavioral aging highlights some unexpected features of complex systems organization

Ten different behavioral tests were performed on a population of young (n = 20) and aged (n = 20) Fischer 344 rats. The relationship structure among these tests was studied by principal component analysis applied both to the entire data set and separately to the two age groups. This analysis proved very useful in highlighting a global index of the rat "behavioral" age based on the entire test set. The analysis effected separately on the two age groups evidenced qualitative differences between them that were linked to the different meaning the same test would assume in rats of different ages. From an overall methodological viewpoint, this work indicates that the correlations among behavioral parameters appear to depend on the observational scale and that the spin-glass model represents an appropriate metaphor to approach the study of the correlations in biological systems.