Further Integration Of Emotions Into An Agent's Decision Making Process: A Proposal For A Modular Emotion Integration Architecture For The Purpose Of Model Comparison And Flexibility
Summary
What role emotions play in the decision making process of humans, and what constitutes an emotion in the first place, is not yet fully understood, and how to properly model emotions is therefore also not yet known. In this thesis, I propose a Modular Emotion Integration Architecture (MEIA). Firstly, so that various ways of modelling emotions can be compared for scientific purposes. Secondly, so that concrete implementations that require emotional cognitive agents (for the purpose of realistic behaviour) have a flexible way to model various emotional agents, to arrive at more desirable behaviour. I did this by first outlining a general agent architecture, so that I could clearly describe how to integrate emotions without committing to any one agent architecture. I then outlined, from various emotion theories and studies, the different factors that go into various emotions (like causes and effects), to allow an as modular as possible construction of emotion models. I concluded by giving some concrete examples of how to use MEIA to apply different emotion models to agents with the same general agent architecture, and how this would affect behaviour.