UMass Machine Learning and Friends Lunch | Main / Emotional Cellular Automata For Ecological Modeling

Abstract: Cellular Automata have recently become popular for modeling predator-prey dynamics in ecology. However, models are not necessarily ready to be relied on for predictions of behavior. In our work, we utilize computational models of emotion to develop a new set of cellular automata for modeling predator-prey dynamics. Our stochastic system describes interacting foxes that feed on rabbits that feed on carrots. Emotions enable foxes and rabbits to improve their decisions and share their experiences with neighboring conspecifics. We show that emotions can significantly impact the model dynamics, including initial game theoretic results pointing to the evolutionary advantage of emotions in predators and prey.