Learning to Discriminate Between Multiple Possible Environments: an Imprinting Scenario

This paper presents work extending the evolution of learning using dynamical neural networks. Previous experiments in this domainhave been limited to evolving agents to discriminate between a maximumof 2 different environmental situations. In this paper we extend beyond this for the first time. We use discrimination between as many as 5 different tones in a scenario designed to reflect the basics of parentalimprinting in, for example, birds.