Nature s distributional-learning experiment door