Anemotion model for non-player characters inspired by neurotransmitters

Many video games depend solely on pre-written scripts, limiting character interaction complexity to what the writers and developers can explicitly design and implement. Nevertheless, real-life decision-making involves various human elements, particularly emotional relationships, which evolve unpredictably. One way to overcome the restrictions of scripted actions is to have non-player characters capable of simulating human emotions. However, the emotion models found in the literature do not adequately represent the dynamic and complex emotions that arise when an NPC is affected by an action. The Lövheim model has the advantage of associating emotions with the effects of neurotransmitters, which are directly related to what actions provoke in people. However, that model is complicated to associate with the diversity of emotions an NPC should experience in a game. In this paper, we propose a new emotion model for NPCs that combines the dynamics of the Lövheim model with the richness of Plutchik’s wheel of emotions, which we call the Extended Lövheim Cube. The combination of emotions caused by neurotransmitter variations must consider the emotional states’ fuzziness, which we consider using fuzzy logic in our model. We validate our Extended Lövheim Cube model by imposing typical in-game actions on an NPC and comparing the resultant emotions with expected ones. Our results show that the fuzzy logic was to able to keep the desired features while having acceptable accuracy values.