Simple Controllers Select for Soft Modes

All biological systems must respond to environmental stresses and perturbations, and often do so through pathways that integrate many kinds of information about their environment. How can a simple controller maintain a complex, high dimensional system at homeostasis? We find that evolving a system for homeostasis in uncertain, high dimensional fluctuating environments leads the system to develop a slow dynamical mode, if the controller constrained to be 'simple' but not otherwise. When constrained to be simple, our theory also predicts that controllers evolved to buffer environmental perturbations will also buffer mutational perturbations. We show evidence of this duality between environmental and mutational robustness in experimental data from yeast knockout experiments. Using transcriptomics data from yeast, we also test our theory's counterintuitive prediction that knocking out a simpler controller effectively decreases the dimensionality of the impact of environmental perturbations. Our work suggests one possible origin of slow modes in evolved biological systems with implications ranging from cryptic genetic variation to global epistasis.