| The field of evolutionary computing includes a variety of simulation algorithms and problem-solving techniques, such as genetic algorithms and genetic programming, that use abstract versions of principles and mechanisms from biological evolution. There still remain many ideas, principles, and mechanisms from evolution, including molecular evolution and evolutionary developmental biology, that await adoption or further exploration in evolutionary computing. Many principles and mechanisms are already widely adopted and developed, such as mutation, recombination, selection, adaptation, and co-evolution, but plenty more that are known to contribute to the power of evolution in nature have barely begun to be explored, developed, and harnessed.;One such mechanism is that of exaptation. Exaptations occur when a trait is co-opted to serve a new role that it previously had not. Exaptation in nature is a valuable source of variation and innovation that augments the variation and innovation generated by mutation at the genetic level. When exaptations occur, complete 'fully-evolved' structures, behaviors, and entities are made available to serve or enhance a new function without having to be built up from scratch. Side-effects of old traits can be amplified by selection, enriching the repertoire of abilities for an evolutionary lineage of organisms.;As processing power continues to increase, evolutionary computing systems will be able to take steps that bring them closer to the levels of complexity and power found in biological evolution. Better study and development, and more widespread adoption of the mechanisms that drive natural evolution will need to take place to help realize the full potential of evolutionary computing in the coming years.;This work examines the current use of exaptation in evolutionary computing, proposes a framework of categorization for different types of exaptation and functional shift mechanisms, and develops a number of example exaptive systems to test and demonstrate the utility of exaptation in a computational context. |
