A Story-Based Phenotype for Self-Interested Affective Agents

摘要

Typical strong models of computational entities contain mentalistic notions that are, based upon psychological models of cognition. However, these cognitive models are continuously evolving and consequently, the strong computational models lag behind the cognitive models. Our approach to address this problem is to base our strong model on a story superstructure of cognition which specifies the most important capabilities for entities. The positioning of computational entities to represent characters in stories has two consequences. Firstly, the attributes of our entities need to be specified by analysing the requirements for typical situations in stories in which characters interact. Secondly, the user of the computational environment which contains our entities is positioned as an author with an active deterministic influence in the environment. These theories are tested in a practical demonstration, where our story construction methods to generate and guide specific computational entities, whose situated interaction produce a story from the point of view of each individual. Application areas for our story-based computational entities include psychological and ecological modelling, and in generating realistic stories for mass entertainment.