Krizhevsky et al. observed the phenomenon we call branch specialization in the first layer of AlexNet by visualizing their weights to RGB channels; here, we use feature visualization to show how this phenomenon extends to the second layer of each branch. For example, the branch specialization we observed in AlexNet – the first layer specializing into a black-and-white Gabor branch vs. a low-frequency color branch – is a surprisingly robust phenomenon. For example, here we see a high-frequency black-and-white branch, a mid-frequency mostly black-and-white branch, a mid-frequency color branch, and a low-frequency color branch.