implementation, the system can still reason about its use of different mind dynamics by considering each MindAgent as a virtual Procedure with a real SchemaNode attached to it. This can lead to some valuable learning, with the obvious limitation that in this approach the system is thinking about its MindAgent