Claes Strannegård: Ecosystems of artificial animals

I will present a computational model of ecosystems that develop over time. An ecosystem is modeled as a block's world, e.g. a Minecraft world, together with a population of artificial animals (animats) that live in this world and move, eat, learn, make decisions, interact with other animats, reproduce, and die. The ecosystem develops from an initial configuration according to a small set of laws of physics and biology. The population undergoes genotype development (addition and removal of animats) as well as phenotype development (learning and forgetting at the individual level). Each animat has its own sets of needs, sensors, and motors. It also has its own ever-changing memory that it uses for decision making. The mechanisms for genotype and phenotype development are generic, i.e. the same for all animats. They are regulated by parameters that are encoded in the genome of each animat. In the case of learning and decision making, they are also regulated by the status of the animat’s needs. I will give examples of ecosystems in which intelligent behavior arises spontaneously as a result of learning and also ecosystems where cooperative behavior arises spontaneously. Finally I will discuss some potential applications of this model to artificial intelligence and population dynamics.