Parallel Distributed Organization of a Memory of Stories

Abstract

Narrative generation is a fundamental issue in seeking a human-like artificial intelligence. The basic objective of this study is to elucidate the computational principle of narrative generation in minds and societies. In this paper, I propose a memory system for the basis of story generation as an internal process of a cognitive system or an artificial mind. The proposed memory system is designed as a parallel distributed system. Each of memory items—stories, events, concepts, and schemas—is an internal agent that operates automatically. Through reading narrative texts, connections and spreading activation among memory items are produced via their parallel distributed operations. I present a proof-of-concept implementation of this idea and show a preliminary analysis of the system’s behavior.