Modeling Cognitive Processes with the Recommendation Architecture

The design of an electronic system with the recommendation functional architecture is described and the results of system learning with inputs simulating visual and verbal sensory inputs are presented. The differences from other cognitive architectural approaches are discussed, and it is emphasized that the recommendation architecture makes it possible for a system performing a complex combination of functions to be constructed, to recover from failures and damage, and to learn without affecting earlier learning. The properties of the implemented system supporting this conclusion are described. Given the understanding of the capabilities of the recommendation architecture derived from experience with the implemented system, complex cognitive processes can be realistically modeled.