(Copyright © 2006 Piero Scaruffi | Legal restrictions - Termini d'uso )
The book describes in detail an architecture (SOAR) for general intelligence.
The universal weak method is an organizational framework whereby
knowledge determines the weak methods employed to solve the problem, i.e.
knowledge controls the behavior of the rational agent.
Universal subgoaling is a scheme whereby goals can be created automatically
to deal with the difficulties that the rational agent encounters during
The engine of the architecture is driven by production rules that fire in parallel and represent task-dependent knowledge. The architecture maintains a context which is made of four slots: goal, problem space, state and operator. A fixed set of production rules determines which objects have to become current, i.e. fill those slots. In other words, they determine the strategic choices to be made after each round of parallel processing.
A model of practice is developed based on the concept of chunking, which is meant to produce the power law of practice that characterizes the improvements in human performance during practice at a given skill. Rosenbloom describes the XAPS3 system, which was designed to model goal hierarchies and chunking. Each task has a goal hierarchy. When a goal is successfully completed, a chunk that represent the results of the task is created. In the next instance of the goal, the system will not need to fully process it as the chunk already contains the solution. The process of chunking proceeds bottom-up in the goal hierarchy. The process of chunking eventually leads to a chunk for the top-level goal for every situation that it can encounter.
TM, ®, Copyright © 2006 Piero Scaruffi