Let's clarify something... when dealing with antenna effect there are not much layers involved, just POLY, M1 and (if you process have) M2 (or top, depends on your process naming convention...)
The charge you might have trapped when processing POLY mask should be having a discharge path as soon as you process M1 (by connecting that poly with a diffusion that could be a transistor drain, or whatever, or an antenna diode).
When you process the M1, additional charges will be present to the already caused by POLY stage (if a discharge path is not present). There is when you do (with a shortest M1 as possible) a layer hopping in order to minimize the additional charges accumulation. In other words, an as I stated in other posts, this will allow you to avoid getting a violation error when checking (since the check is based on area ratios) but, as you can see, this is only a partial solution... but this answers why you do an upper layer hop.