I stumbled across this article recently about how current theories of perception are similar to what magicians have been exploiting for years. Essentially, the idea is that we can only “see” what our attention is focused on at a given time. They liken it to a “spotlight” which roams around and only let’s you take in certain things under its view. Though in the case of vision all the things out of the “spotlight” are still within your visual field. You just don’t “see” them.
As I discuss in this video, panels in the visual language used in comics serve to facilitate this same sort of focusing of attention. Most of the time though, panels serve to exclude all relevant information except for the elements that need to be focused on, or at least clearly distinguish what is relevant from irrelevant. This lets panels provide a graphic manifestation of this mental “spotlight,” allowing the author to control that attention instead of the reader’s wandering eyes (which is one of the reason’s I formally call panels “Attention Units”).
This ties into the argument for why you don’t want to overload a panel with too much stuff, because it becomes too hard to disentangle the attentionally important from unimportant elements. (If you still want to pack info in, inset panels help facilitate this honing of attention).
Even more, when you have too much in several panels sequentially, it becomes too difficult to track all the changes and carry-overs from one panel to another. This is what gives way to things like “parallel cutting.” By switching back and forth between two (or more) scenes, you can highlight the individual aspects of each in panels without risking it becoming overlooked for other information or overloading the system. Of course, doing so introduces other processing demands on the visual grammar, but at least your attention is focused exactly on what is intended to be conveyed.