> I suppose because they resemble the abstractions that make complex language possible

As in models: All analogies are "wrong", some analogies are useful.