Action Progress Organizes Motor Memory

How humans reuse learned movements across different speeds, sizes, and contexts is poorly understood. Conventional theories assume that motor memories are linked to limb states; however, they fail to explain why the structure of skilled actions is preserved when scaled. Here, we propose the novel hypothesis that the brain organizes motor memory according to the relative progress within an action, where the movement is in its progression from start to finish. We tested this hypothesis using three experimental tasks involving single reaches, two-step reach sequences, and button-reach-button sequences. Across all tasks, behavior was more accurately predicted by representations anchored to the relative progress of the overall action than by state variables. Taken together, our findings support a scalable encoding scheme for abstract motor memory. This perspective provides a mechanistic basis for flexible, scale-invariant reuse of motor memories and the remarkable generalization capacity of human motor control.