Mental stress is a global epidemic that can have serious health consequences including cardiovascular diseases and diabetes. Several techniques are available to teach stress self-regulation skills including therapy, meditation, deep breathing, and biofeedback. While effective, these methods suffer from high drop-outs due to the monotonic nature of the exercises and are generally practiced in quiet relaxed environment, which may not transfer to real-world scenarios. To address these issues, this dissertation presents a novel intervention for stress training using games and wearable sensors. The approach consists of monitoring the user's physiological signals during gameplay, mapping them into estimates of stress levels, and adapting the game in a way that promotes states of low arousal. This approach offers two key advantages. First, it allows users to focus on the gameplay rather than on monitoring their physiological signals, which makes the training far more engaging. More importantly, it teaches users to self-regulate their stress response, while performing a task designed to increase arousal. Within this broad framework, this dissertation studies three specific problems. First, the dissertation evaluates three physiological signals (breathing rate, heart rate variability, and electrodermal activity) that span across the dimensions of degrees of selectivity in measuring arousal and voluntary control in their effectiveness in lowering arousal. This will identify the signal appropriate for game based stress training and the associated bio-signal processing techniques for real-time arousal estimation. Second, this dissertation investigates different methods of biofeedback presentation e.g. visual feedback and game adaptation during gameplay. Selection of appropriate biofeedback mechanism is critical since it provides the necessary information to improve the perception of visceral states (e.g. stress) to the user. Furthermore, these modalities facilitate skill acquisition in distinct ways (i.e., top-down and bottom-up learning) and influence retention of skills. Third, this dissertation studies reinforcement scheduling in a game and its effect on skill learning and retention. A reinforcement schedule determines which occurrences of the target response are reinforced. This study focuses on continuous and partial reinforcement schedules in GBF and their effect on resistance to extinction (i.e. ability to retain learned skills) after the biofeedback is removed. The main contribution of this dissertation is in demonstrating that stress self-regulation training can be embedded in videogames and help individuals develop more adaptive responses to reduce physiological stress encountered both at home and work.
