Memory System and Memory Types for Real-Time Reasoning Systems

Peter Isaev, Patrick Hammer

Abstract: In this paper we discuss different types of memory from several cognitive architectures in the context of Artificial General Intelligence. We then introduce the memory system for the Artificial General Intelligence system based on NARS with a description of its related features. Then we identify and characterize NARS memory into different types in terms of use, duration (short and long-term) and type (procedural, episodic, declarative, etc.). At the end we also provide demonstration of memory functionality showing how the same piece of knowledge can contain declarative, episodic and procedural components within it.