Weinreich, D. M.; Sgouros, T.; Raynes, Y.; Burtsev, H.; Chang, E.; Rajakumar, S.; Bravo, I. G.; Petak, C.

Noise is intrinsic to information transmission, and information transmission is intrinsic to life. Critically, while biological noise is random, its amount is in part genetically determined. Here, we explore the population genetics of alleles that heritably influence the amount of noise in any biological mechanism. We argue that biological noise is a double-edged sword: almost always inducing deleterious phenotypes but also occasionally yielding high fitness outcomes. We further suggest that this implies the existence of an equilibrium amount of noise, at which the advantage of producing additional, rare beneficial phenotypes is balanced by the cost of producing the vastly more common deleterious phenotypes. The location of this equilibrium will reflect the rate at which its environment is changing, the persistence of association between noise-modifying alleles and the phenotypic perturbations they induce, the mode of selection, and the population size. The population genetics of alleles that heritable change mean trait values have been established for nearly a century, but to our knowledge, this is the first general treatment of the evolution of heritable changes in the statistical distribution around that mean. Our framework generalizes modifier theory in the sense used by Altenberg, Feldman and others, and resolves teleological criticisms of the hypothesis that evolvability can evolve via natural selection. We conclude with an outline of open theoretical and empirical questions posed by our framework.