Abstract
<p>Video games, by providing dense practice of goal-directed motor actions with immediate and salient feedback, have drawn growing interest in motor learning and rehabilitation research. Yet most studies do not specify which game design mechanisms drive improvement—a gap that may hinder our ability to understand and exploit the unique potential of video games for motor learning.To address this, we propose a refined taxonomy of adaptive difficulty and reinforcement (AD&amp;R) schemes comprising three mechanisms: Skill-Based Progression (SBP; the incremental adjustment of game difficulty), Target Manipulation Boosting (TMB; powering-up the player's impact on targets), and Control Boosting (CB; amplifying virtual movement relative to physical input). We discuss the neuroscientific features of these distinct schemes in the context of reinforcement learning.Following PRISMA 2020 guidelines, we systematically reviewed the prevalence of these schemes across commercial video games (n=162), clinical studies (n=287), and non-clinical studies (n=98) published between 2020 and 2024. Using a rigorously validated Large Language Model coding pipeline, our analysis (n=547) revealed that all three AD&amp;R schemes are underrepresented in games employed in academic research compared to commercial games. Specifically, TMB and CB are nearly absent in research, showing 67.9% prevalence in commercial games compared to 1.4% in clinical and 2.1% in non-clinical games for TMB, and 58.6% compared to 1.4% and 3.1% for CB.Ultimately, this study demonstrates that academic research paradigms significantly underutilize performance-amplifying game mechanics. Integrating TMB and CB into neurorehabilitation protocols provides an actionable framework to optimize motor skill transfer, emotional engagement, and user agency.</p>