IRCS Technical Reports Series

Document Type

Technical Report

Date of this Version

December 1994


University of Pennsylvania Institute for Research in Cognitive Science Technical Report No. IRCS-94-33.


In this paper we review Subsumption Architecture and Discrete Event Systems. These approaches present diverse methodologies for dealing with control of interactions. They often take diametrically opposite directions in addressing specific issues. Subsumption architecture expects limited knowledge of the environment, no explicit representation, limited reasoning capabilities and no centralized control. At the other extreme lies Discrete Event Systems, which require, at least in manufacturing and communication: a well-structured environment; explicit representations and models; and have limited reasoning capabilities and centralized control. Both offer benefits and limitations which should really be evaluated and traded off when attempting to build a system. However, combining aspects from these two approaches will not address and resolve all issues. We conclude that while both approaches are powerful there is more to intelligence than just behavior and control, and discuss the limitations and benefits entailed by both.



Date Posted: 19 September 2006