As I recall from Pysch 101 - taken in the 60s so my recollection may not be exact - human beings think in pictures and images rather than words. How the psychologists figured that out I do not know. Words, both verbal and written, are our attempts to describe those pictures in our heads. Using diagrams, drawings, models, etc. gets us even closer to the pictures in our heads. However, we also do not think statically. Our pictorial thoughts are in association with other images in our head, and tend to be more dynamic, that is moving. To accommodate this, we develop prototypes that show the data in conjunction with the interface in conjunction with the process of using both. Now our model gets really close to what we see in our heads. Working with information systems, however, the picture we have in our heads - at least in the users' heads - is somewhat foggy or vague since it deals with abstract and future reality, and really doesn't clear up until we see the prototype in action.
Now here's the kicker. The sufficiency bias says that we will be satisfied with something that satisfies our specific needs rather than to hold out for something better. This means that what we are looking at now becomes the picture of what we want in our heads. Later, we might see something else that changes the picture and expect the delivered system to have that something else. And round we go.
Isn't inexact science fun?