Programming with Affect: How Behaviour Trees and a Lightweight Cognitive Architecture Enable the Development of Non-Player Characters with Emotions

Abstract

There is little support in game development tools for implementing emotion and mood in non-player characters. In this paper, we present a tool that integrates behaviour trees into an emotion modeling framework, based on ALMA, to add emotion and mood to NPCs. We introduce two new nodes to traditional behaviour trees, the Emotion Adder, which triggers emotions in NPCs and the E-Selector, which incorporates the NPC’s mood in the decision-making process. This EMoBeT framework arms developers with a tool that will assist them in integrating and manipulating psychologically valid moods and emotions of NPCs using the familiar behaviour tree model.