About reasonable structures

[junseokoh1]

Hi, thank you for your nice work and paper.
I'm quite interested in you work and I have some question.
In paper, qualitative result show reasonable structures. There is no image that structure collapses and the meaning of the object disappears like a five-legged zebra

I wonder where does reasonable structure comes from. I think learning internal distribution of patch with proper receptive field at a single scale doesn't guarantee reasonable structure.

I will be appreciated for your reply, thank you very much!
Best wishes