Evolving Group Behavior

Abstract

In this report we present our work on group behavior through an evolutionary algorithm. In nature several factors have been proposed as being responsible for the evolution of group forming, including the topics of our study: food distribution and the ability to perceive information about other creatures. Two research questions were addressed: what is the effect of food clustering on group behavior and what is the effect of the ability to perceive the success in food gathering of other creatures on group behavior. Two artificial life simulations were implemented. The first simulation evolved artificial creatures in environments of different food clustering, the second evolved creatures with and without sensors to perceive information about the food gathering success of other creatures. The creatures of the last generations were analyzed by measuring the amount of time they spend in each others smell radius during a test condition. The use of the success sensor was measured by testing for group behavior against creatures of different success. It was found that both higher food clustering and the ability to perceive food gathering success of others lead to more group behavior.