Selection-based estimates of complexity unravel some mechanisms and selective pressures underlying the evolution of complexity in artificial networks

Using artificial networks and different estimates of complexity, this chapter discusses the biological meaning of the different estimates of complexity: while informational complexity (IC) reflects more the complicatedness of the implementation, phenotypic complexity reflects more the complexity of the task performed. The estimation methods of phenotypic complexity offer, therefore, a relevant framework to study complexity and its evolution. It suggests that complexity may be better understood through the interaction of the organisms and its selective environment. These methods are applied to more biological networks to try to uncover some real molecular determinants of complexity and to develop some further less-constrains models in which the size of the networks is free to evolve and the phenotypes used to infer fitness are less fixed. Finally combining these approaches with some topological estimates of network complexity may be an interesting perspective to understand the topological organizations that promote phenotypic complexity.