Varying Environments Can Speed Up Evolution
31 Jul, 2007 11:28 am
Nadav Kashtan is a specialist of evolution computer simulations. He authored a study on the subject published in the Proceedings of the National Academy of Sciences.
Scientists use computers to evolve artificial systems, which serve as an analogy to biological systems. The simulations mimic natural evolution by incorporating three fundamentals of evolution: replication, variation (e.g. mutation and recombination) and selection. In the simulations population of genomes (which are represented by strings of bits in the computer evolve towards a given goal. The goal can be a computational goal: to compute a desired output on given inputs. A genome represents a “digital organism”, a computational network for example (such as a neural network or any other computational model). Each of the genomes is assigned a fitness based on its performance: how well it achieves the goal. Genomes with higher fitness are given a higher probability to replicate. Genetic operators such as mutations are then applied to alter the genomes of the next generation. As generations proceed the fitness of the genomes increase, until the goal is achieved.
Studying natural evolution is made difficult by incomplete knowledge of genetics, the extinction of prehistoric life forms, and lack of information on the dynamic environment through Earth history. Thus, computer simulations can serve as an alternative way to study evolution and help to overcome the mentioned difficulties.
Usually these computer programs take a very long time to reach their goal. How
did you manage to speed up evolution?
Typically, when you simulate evolution towards a constant fixed goal, a logarithmic slowdown in evolution is observed: longer and longer periods are required for successive improvements in fitness. Simulations can take many thousands of generations to reach even relatively simple goals such as Boolean functions of several variables. When we simulated evolution towards goals that vary over time, we surprisingly found out that in some cases evolution was considerably faster - each of the goals achieved consistently in fewer generations. The highest speedup was observed when the goals changed in a modular fashion, where each new goal shared some of the subgoals with the previous goal. Interestingly, we found out that the harder the goal the larger the speedup.
What new information do you expect to find out about evolution?
A central question in evolution is how we can explain the speed at which the present complexity of life arose. To understand the speed of natural evolution, it is of interest to find generic ways, compatible with natural conditions, in which evolution in simulations can be sped up. Our finding shows that varying environments can speed up evolution. This suggests that varying environments may contribute to speed up natural evolution. Although we aimed at understanding the speed of biological evolution, it may also apply to evolutionary approaches in engineering and optimization algorithms.
Interview by: Clémentine Fullias
Reference: Kashtan, Nadav, et al. 06-11630: "Varying environments can speed up evolution," Proceedings of the National Academy of Sciences 2007.
