I plan to continue and generalize this work, or something closely related, for my Ph.D. work. For example, I would like to explore the design of reflective systems that can reason effectively about their own past and future computational behavior, including their reflective behavior, and act rationally to change what their future behavior will be. The ultimate goal would be a system that knew enough about itself, and about programming, that it could adaptively refine or restructure parts of itself, with the aim of improving its own performance. Such a system would greatly facilitate further development of complex AI programs.