> It's reasonable and natural to have a mental model that the control moves the observer. (move a control up to aim your eyes up)

> It's also reasonable and natural to have a mental model that the control moves the object. (move a control down to "grab" the object and move it down)

Except that neither of those is the reason you'd want inverted controls. You want inverted controls because you have to lean back to look up. The model is that the control moves you.