The team built tiny, simple robots, just a few centimeters tall, made of wheels for mobility and a basic sensory system equipped with a camera. Moving around an arena, they would seek out “food”–small discs scattered by the researchers. To lend a biological flavor to the arena, the team programmed each robot with stream of ones and zeroes that acted as a sort of digital genome. The researchers found it most practical to then proceed with computerized simulations of the behavior of the actual, physical robots. (They periodically compared the simulated robots’ behavior with that of the physical robots; the comparisons checked out, Floreano told ScienceNow.) The team then introduced a new rule: robots were allowed to share their food with one another, to help ensure that one of their robo-brethren survived to the next generation in lean times. Running the virtual robots through hundreds of generations, they discovered something remarkable: the robots behaved just as Hamilton had predicted species would. Altruism essentially “evolved” among the robots–and when the robots had their digital genome coded to make them closer kin, they evolved altruism all the more rapidly. (via Can Robots Really Teach us About Altruism? – Technology Review)