Physiological DDA: Physiological Sensor-based Dynamic Difficulty Adjustment for Enhanced Video Game Engagement

Abstract

Dynamic Difficulty Adjustment (DDA) systems aim to maintain player engagement by adapting game challenge to match one's skill level. However, current approaches primarily rely on performance metrics that capture in-game metrics or transactional data rather than underlying player states. This paper introduces an approach that simultaneously assesses challenge and engagement from physiological responses to enable direct difficulty adjustment based on player engagement. We developed a hybrid LSTM model that predicts challenge and engagement from multimodal physiological signals (N=10), including eye tracking and electrodermal activity. We then evaluated the player experience of physiological DDA against single-level, gradual increase, and performance DDA methods (N=12). Results showed that physiological DDA significantly increased player engagement compared to single-level and gradual increase conditions, while enhancing cognitive challenge without increasing workload. These findings suggest that DDA based on direct assessment of challenge and engagement can help players stay in flow without relying on game-specific performance metrics.