Operator Subgoaling in Soar
Operator Subgoaling in Soar
In Soar, deliberate goals take the form of operators, which are often
represented declaratively in other cognitive architectures. When a
operator is selected in Soar, the resulting state from the application
of the operator is built by the successive firings of productions during elaboration. This is like
representing an operator as a fixed piece of code. However, if the
operator implementation is not known directly, an impasse results and a subgoal is
created to implement the operator. This process is known as
operator subgoaling. The operator may be implemented within
the subgoal by drawing on some declarative knowledge of how to change
the state or decomposing the operator into a set of smaller operators.
The advantage of operator subgoaling is that the implementation of
operators is chunked over time (since
learning is continuous and reflexive) and different
instantiations of the same operator may be built for different
situations.
Return to the top of this architecture.
Current Location: Soar-Capabilities-Universal Weak Methods-Operator Subgoaling