Chaining Distinct Tasks Drives the Evolution of Modularity

Brett Calcott
2014 Artificial Life 14: Proceedings of the Fourteenth International Conference on the Synthesis and Simulation of Living Systems   unpublished
I introduce a novel method for evolving modularity in gene regulatory networks. Like previous models of modular evolution, it relies on selecting networks to perform a task consisting of distinct sub-tasks, with the aim of producing modules that perform these sub-tasks. Whilst existing models structure these sub-tasks in parallel and then combine the output, this model chains sub-tasks together, so they must be performed one after another. This task structure resembles the selective constraints
more » ... lective constraints undergone in multicellular evolution, where genetic networks must (a) integrate multiple cues to establish what environment they are in, and (b) express a pattern of gene activity on the basis of this environment. I show that the modules produced in these networks exhibit the hallmarks of modularity: existing modules can change independently of others, and modules can also be re-used and combined in various ways.
doi:10.7551/978-0-262-32621-6-ch112 fatcat:2y7mqeaufbajfjs4amivodv3xq