Measures of useful information in layered networks

Phil Fraundorf, Dept. of Physics & Astronomy, University of Missouri-StL,

Measures of useful information based on Kullback-Leibler divergence (like available work in thermodynamics, mutual information in correlation theory, and Akaike information criterion in model selection) are typically applied on one level of organization. We show here how their application to layered subsystem-correlation models opens new doors in the study and maintenance of emergent complexity. In particular, a simplex model of niche-layer multiplicity in metazoan communities shows promise for tracking community health and the effects of stress, as well as for modeling the impact of idea-code expression on community dynamics. The value of informing code expression to correlations on multiple scales is already well illustrated by the success of eukaryotes at forming metazoan individuals.