Spatial mappings are essentially how I express memories that others seem to hold as mental images of some kind, many of those _essential to function_, like "how do I drive from A to B", "how do I accomplish this complex task", "how do I recognize a place I've been before" (that last one is.. an interesting issue, because it turns out the things I use to recognize a scene are _almost_ sufficiently unique, but _not quite_; I occasionally get locations confused entirely because their streets meet at the same angle/distances, or the hedge and building are similar heights).
More likely than not there are tradeoffs to using strictly visual vs. visual+ vs. strictly non-visual mental models in all domains, and there is no universally optimal strategy (and that teams of mixed cognitive styles are best overall). You know, no free lunch.