Cerebellar activity is triggered by reach endpoint during learning of a complex locomotor task

Locomotion in complex environments requires precise timing and control of paw movements to adapt steps and coordinate limbs. The cerebellum is thought to support moment-to-moment prediction and correction of these movements, but how its cortical neurons encode paw dynamics during locomotion requiring step-specific adjustments remains unclear. We developed LocoReach, a task in which mice walk on a runged treadmill, requiring each step to reach for the next rung. With training, mice improved performance, showing fewer missteps, longer strides, and faster swings. Optogenetic disruption of cerebellar activity shortened swing duration, highlighting the online contribution of lobule simplex. Electrophysiological recordings revealed that molecular layer interneurons and Purkinje cells exhibited activity changes around swing-stance transitions, primarily for the ipsilateral paw but also across multiple paws. Notably, interneurons showed larger responses during longer strides acquired through learning. These findings indicate that cerebellar activity is precisely aligned to step-cycle events, supporting adaptive locomotor control. How cerebellar circuits control precise paw placement during locomotion remains unclear. Here, the authors show that cerebellar neurons encode paw-specific step transitions and adjust their activity with learning, supporting precise motor control.