People using brain-computer interface are more efficient when both human and machine are allowed to learn. Researchers trained two tetraplegic users to compete in the international Cybathlon BCI race. Both incrementally learned how to control the BCI, and obtained the best performances.