Inheriting Constraint in Hybrid Cognitive Architectures: Applying the EASE Architecture to Performance and Learning in a Simplified Air-Traffic Control Task

摘要

This report describes the development and evaluation of models of behavior in the AMBR air-traffic control (ATC) and category learning task. We adopt, as constraints, several of Newell's principles and recommendations on the development and use of models and architectures. The EASE architecture, developed in the course of the AMBR project, came about as a result of the constraints of 'listening to the architecture' and identifying areas where the set of mechanisms of the original architecture (Soar) needed to be amended. This work also reuses existing Soar models of category learning (SCA), and operationalizes and expands a non-architectural model found in the literature (RULEX).