A Computational Model of Suspense for Non-Narrative Gameplay

In recent years, Various mechanisms have been proposed to optimize for players’ emotional experience. In this paper, we focus on suspense, one of the key emotions in gameplay. Most previous research on suspense management in games focused on narratives. Instead, we propose a new computational model of Suspense for Non-Narrative Gameplay (SNNG). SNNG is built around a Player Suspense Model (PSM) with three key factors: hope, fear, and uncertainty. These three factors are modeled as three sensors that can be triggered by particular game objects (e.g., NPCs) and game mechanics (e.g., health). A player’s feeling of suspense can be adjusted by altering the level of hope, fear, and uncertainty. Therefore, an SNNG-enhanced game engine could manage a player’s level of suspense by adding or removing game objects, diverting NPCs, adjusting game mechanics, and giving or withholding information. We tested our model by integrating SNNG into a Pacman game. Our preliminary experiment with nine subjects was encouraging.