Video game practice is associated with enhanced value-guided exploitation under probabilistic uncertainty

This study examined whether habitual video game play influences reinforcement learning dynamics, feedback adaptation, consolidation, and motivational biases. Two groups of participants (gamers and controls) completed a Probabilistic Selection Task that assessed learning from positive and negative feedback across three phases: Learning, Test, and Transfer. Mixed-effects modeling revealed that gamers showed enhanced learning trajectories, particularly under high-uncertainty conditions; however, computational modeling indicated no differences in learning rate (α), suggesting comparable value updating across groups. Gamers exhibited a higher tendency toward model-based exploitative (value-consistent) choices during learning compared to controls. In the Test phase, gamers demonstrated higher accuracy, especially on difficult stimulus pairs, suggesting more effective use of learned value representations under no-feedback conditions. This was further supported by greater model-based exploitative choices for challenging pairs. However, while no group differences emerged in transfer-phase approach/avoidance biases (VLBI), gamers showed greater decision consistency (higher inverse temperature, β) and increased exploitative choices in high-value (A-present) trials, indicating more deterministic value-guided behavior during generalization. These findings suggest that habitual video game play improves how efficiently learned values are translated into action under uncertainty, highlighting the potential of game-like environments to enhance value-guided decision-making and adaptive behavior in educational and clinical settings.