LEARNING AND EVOLUTION IN NEURAL NETWORKS

This article describes simulations on populations of neural networks t hat both evolve at the population level and learn at the individual le vel. Unlike other simulations, the evolutionary task (finding food in the environment) and the learning task (predicting the next position o f food on the basis of present position and planned network's movement ) are different tasks. In these conditions, learning influences evolut ion (without Lamarckian inheritance of learned weight changes) and evo lution influences learning. Average but not peak fitness has a better evolutionary growth with learning than without learning. After the ini tial generations, individuals that learn to predict during life also i mprove their food-finding ability during life. Furthermore, individual s that inherit an innate capacity to find food also inherit an innate predisposition to learn to predict the sensory consequences of their m ovements. They do not predict better at birth, but they do learn to pr edict better than individuals of the initial generation given the same learning experience. The results are interpreted in terms of a notion of dynamic correlation between the fitness surface and the learning s urface. Evolution succeeds in finding both individuals that have high fitness and individuals that, although they do not have high fitness a t birth, end up with high fitness because they learn to predict.