The Cognitive Cost of Non-Player Character Companions in Video Games

<p dir="ltr">The personalisation of video games allowing the catering of individual player needs and preferences, with approaches ranging from character customisation to narratives evolving based on player choices. These traditional methods rely on explicit player input. A promising alternative is adaptation, which has emerged as a frontier for creating immersive and tailored gaming experiences. Enabled by advancements in wearable technology, adaptation can now include the use of non-intrusive hardware and machine learning to track subtle changes in player experience and cognitive state over time, allowing real-time changes of game content to occur. </p><p dir="ltr">Non-player character (NPC) companions are a emerging avenue for implementing adaptation, given their integral role in shaping the player experience. This thesis explores how adaptive NPC companions can enhance personalisation by utilising psychophysiological data. However, current methods for designing adaptive characters lack standardisation, and there is limited validation of the data used for adaptation. To address these gaps, this thesis proposes a standardised framework for creating adaptive companions and applies it in two user experiments. The first experiment examines methods for measuring cognitive load during gameplay as a proxy for player engagement, while the second employs a Wizard-of-Oz methodology to implement and test an adaptive NPC companion using previously validated data types. </p><p dir="ltr">The findings indicate that adaptive companions differentially impact the player experience based on player skill, highlighting the potential of this approach to meet diverse player needs and preferences. This work provides evidence for the importance of continuing research into adaptive gaming systems to further advance personalised and engaging video game experiences.</p>