R1 and R2 correctly point out that relating intermediate representations in neural networks to brain