Predictive use of environmental regularities requires action relevance

Efficient navigation often depends on the ability to learn and exploit environmental regularities through predictive motor adjustments. We conducted two virtual reality experiments in which participants walked through a museum corridor towards one of two exit doors. In Experiment 1, a moveable obstacle appeared with high probability on one side depending on the block, but participants could delay their pathway choice until the obstacle became visible. Many adopted a waiting strategy, walking straight until the obstacle appeared and only then adjusting their route. This raised the question of whether this reflected a failure to learn or a strategic delay. In Experiment 2, we introduced a larger static obstacle that forced participants to commit earlier, thereby increasing the cost of late corrections. Under these constraints, nearly all participants consistently selected the correct pathway in anticipation, indicating that the spatial structure could be learned and exploited when early commitment was required. Together, these findings suggest that predictive motor planning does not simply arise from the presence of environmental regularities but reflects an adaptive process of embodied decision-making shaped by task demands, environmental structure, individual strategy preferences, and learning capabilities.