Shifting Evolution Up A Gear

By Will Parker on August 28, 2007 in News

Research appearing in the Proceedings of the National Academy of Sciences suggests that “moving the goalposts” may be one method of speeding-up evolutionary change. Scientists from the Weizmann Institute of Science came to this conclusion after running computer simulations involving “digital genomes” where the simulated goals were continually changed.

Nadav Kashtan, Elad Noor and Prof. Uri Alon of the Institute’s Molecular Cell Biology and Physics of Complex Systems Departments created computer simulations that would mimic natural evolution, allowing them (theoretically) to investigate processes that, in nature, take place over millions of years. In these simulations, a population of digital genomes evolves over time towards a given goal: to maximize fitness under certain conditions. Like living organisms, genomes that are better adapted to their environment may survive to the next generation or reproduce more prolifically.

In nature, evolution takes place under changing environmental conditions, forcing organisms to continually readapt. Achieving even simple goals may take thousands of generations. But when Kashtan, Noor and Alon created a simulation in which the goals changed repeatedly, they found that its evolution actually speeded up. They even found that the more complex the goal – i.e., the more generations needed to reach it under fixed conditions – the faster evolution accelerated in response to changes in that goal.

Interestingly, the scientists found that computerized evolution ran fastest when the changes followed a pattern they believe may be pervasive in nature. In previous research, Kashtan and Alon had shown that evolution may often be modular – involving adjustments to standard parts, rather than wholesale remodeling. They theorized that the forces acting on evolution may be modular as well, and for each goal, they defined subgoals that could each change in relation to the others.

“In an organism, for example, you might classify these subgoals as the need to eat, the need to keep from being eaten, and the need to reproduce. The same subgoals must be fulfilled in each new environment, but there are differences in nuance and combination,” explained Kashtan. “We saw a large speedup, for instance, when we repeatedly exchanged an ‘OR’ for an ‘AND’ in the computer code defining our goals, thus changing the relationship between subgoals.”

Although the main aim of the research was to shed light on theoretical questions of evolution, it may have some practical implications, particularly in computer science, by providing a possible way to accelerate optimization algorithms.