Dual phase evolution enhances adaptation of binary strings

Recent studies have shown that repeated phase changes in large networks (dual phase evolution - DPE) play a role in the evolution of many kinds of systems. However, the contribution of DPE to the origin of modular adaptations remains to be demonstrated. Despite plausible arguments and suggestive evidence from natural systems, a clear proof of the adaptive benefits of DPE in module formation has been lacking. Here, we explore the impact of dual phases on modular adaptation in a simple artificial life model. The results show that DPE can indeed enhance the evolution of modular solutions in complex adaptive systems. Moreover, we show that this effect is enabled not only by the escape from local optima afforded by non-local connections, but also by the distinct and complementary evolutionary dynamics of the two phases.