One problem of increasing significance in computer science is building systems that can sense and learn from their environment. One important example is building an adaptive vision system that can detect, recognize and track objects of interest in its visual field. This project focuses on reinforcement learning, a powerful technique from the field of machine learning, which enables an autonomous active vision system to learn to make eye movements to locations of interest in a visual scene. The goal is to learn the mapping from visual coordinates to the appropriate motor commands for moving the camera from its current point of focus to one in its periphery. As part of this objective, we would like also to explore the use of a predictive world model, as implemented by a neural network, to help the system predict the consequences of its actions.