Abstract
This thesis investigates adaptive game design at the level of player experience, addressing a gap in both procedural content generation and existing adaptive systems, which are typically limited to content creation or difficulty adjustment. It proposes an Adaptive Game Design Framework that dynamically interpolates between designer-defined gameplay presets based on player behavior, consisting of four components: presets, a player profiler, an adaptive core, and an adaptive engine.To evaluate the framework, a prototype top-down dungeon crawler was developed around two contrasting playstyles, action-oriented and stealth-oriented, which define the target experiences of the system. Three gameplay modes were implemented: a static action-focused mode, a static stealth-focused mode, and an adaptive mode that dynamically interpolates between these experiences based on player behavior. A within-subjects study was conducted in which participants experienced all three modes and provided self-reported evaluations based on selected Player Experience Inventory (PXI) dimensions, complemented by behavioral telemetry.
Results show that while no statistically significant differences were found at an aggregate level, the adaptive system demonstrated clear advantages when analyzed relative to individual preferences. The adaptive mode achieved experience ratings comparable to each participant’s preferred static mode while outperforming their non-preferred mode, and produced more consistent responses across participants. These findings indicate that adaptive systems can reduce mismatches between player preferences and gameplay experience.
This research contributes a conceptual and practical framework for extending procedural methods to the level of game design, introducing a parametric approach where gameplay experiences are dynamically adjustable. While exploratory, the study demonstrates the potential of adaptive game design to support diverse player populations and provides a foundation for future research in player-centered interactive systems.