Derandomized Evolution Strategies in Computational Neuroscience

We propose evolutionary “analysis by synthesis” guided by neurobiological knowledge as a powerful tool in computational neuroscience. The challenge is to force artificial evolution to favor solutions that are reasonable from the biological point of view. Such solutions are only likely to evolve if as much neurobiological knowledge as possible is used in the design process. This can be achieved by providing sufficient experimental data to evaluate the evolved systems and by a considerate choice of the basic structure. Additional knowledge can be incorporated into the fitness function and in constraints that ensure biological plausibility.