I could kind of grok XPath without resorting to visualization, but had to visualize trees to cope with XSLT. And for reasons probably anchored in exposure to antique CS texts, I visualize trees upside down (i.e., root at the top). So in this case the monkeys would be rearranging the tree while hanging on for dear life.

Someone, and I wish I'd saved the link, has a visual explanation of XSLT that draws trees from the top down, and color-codes the tree parts targeted and affected by various XPath/XSLT fragments. Just looking at things organized this way was enough to part the clouds enough to give me a grip on the big picture.