Meta-Reasoning in Meta-Reasoning Architectures
Meta-Reasoning in Meta-Reasoning Architectures
Meta-reasoning is the primary capability of MAX. Since most of the
capabilities in a MAX agent are represented as explicit knowledge
(including learning, problem-solving), MAX can reason about when to
learn, what planning method to use, etc. There is not an explicit
meta-reasoning module; rather, by making as much knowledge as possible
available to all behaviors,
meta-reasoning can be done by any behavior. For example, one behavior
could decide that it is time to learn, or time to plan, or whatever.
Meta-reasoning is facilitated by MAX's homogeneous knowledge
representation.
Return to the top of this architecture.
Go to a discussion of this capability
for multiple architectures.
Current Location: Meta-Reasoning-Capabilities-Meta-Reasoning