Dynamic Symbol Systems: An Introduction to the Local Model

Dynamic symbol systems (DSS) are a class of dynamic systems especially developed for modeling situated agents. They combine a symbolic format of representation with a selforganizing dynamics. The theory can be used for theoretical purposes, as a discrete, approximative reconstruction of continuous dynamic systems. It can also be used for the practical design of multi-granular information processing systems, since it is formulated in algorithmic terms. Furthermore, as processing on a conceptual level is concerned, the dynamic system perspective suggests answers to some questions that are hard for logic-oriented approaches. All in all, DSS contributes to bridging the gap between dynamics-oriented, bottom-up techniques, and representation-oriented, top-down perspectives in AI. The paper gives an informal introduction to the part of DSS theory concerned with local phenomena, while the global part of the theory is very briefly sketched. October 1994 draft version submitted to AISB-95