Meta-Reasoning in Soar

Meta-Reasoning in Soar

A Soar agent can do meta-reasoning in some ways, but not in others. For example, when solving a problem, one of the first decisions made is what problem-space to use. This is used to set up bounds for the search, and to specify the problem-solving method. However, if there are several problem-spaces that might apply, an impasse occurs, and problem solving must be used to decide which problem-space to apply.

So this type of meta-reasoning is done in the same manner as other problem-solving, due to the uniform nature of the representation. On the other hand, many aspects of Soar's operation are impenetrable by, or not under the direct conrol of, the problem solver. For example, this means that it is not possible for a Soar agent to reason about whether to build chunks.


Return to the top of this architecture.

Go to a discussion of this capability for multiple architectures.


Current Location: Soar-Capabilities-Meta-Reasoning

Go to NEXT page.