Explication of a passage from Gombrich on pictorial realism

Ernst Gombrich can be a lot more difficult than he at first appears.  I gave a quiz in my Introduction to Aesthetics this week in which I asked students to explicate page 11 of the third edition of Aesthetics:  A Reader in Philosophy of the Arts ed. David Goldblatt and Lee B. Brown.  The passage is from a selection by Gombrich called "The Limits of Likeness" which comes from his book Art as Illusion.  Several students found it quite difficult to do this.  So, as an aid, here is my own take on the passage.  First, in order to properly interpret a piece of scholarly prose like this one one should consider it in context.  Look, in particular, at what came before it.  What comes prior to this passage is the most important part of the reading, the last paragraph of page 10.  There, Gombrich defines a "correct view" [for a drawing or other visual representation] in this way:  "those who understand the notation [i.e. the style used in the drawing] will derive no false information from the drawing."  He allows for different styles to be equally correct.  He further gives what I would call his definition of pictorial realism in terms of what he calls a "complete portrayal."  He is tentative about this:  "a complete portrayal might be the one which gives as much correct information about the spot as we would obtain if we looked at it from the very spot where the artist stood."  As can be seen, this is an information-based account of realism.

Now to look at page 11.   He had said earlier that he was interested in the "riddle of styles."  So, here he is continuing with this.  Consider as examples of styles the style of the German students in his story who drew with great precision using hard pencils, and the style of the French students who drew the same scene using broad brushstrokes.  So, styles differ in that they require different "articulations" and they allow the artists to ask different questions (for example the hard pencil user is going to ask questions about how sharp lines can depict shadows, for example, realistically).  Still, Gombrich argues, the information that we get from a scene is so complex that no one picture and no one style will capture it all.  Some would say that variety in style is a matter of subjectivity of vision:  each person having his own way of seeing.  But it is more a matter of there being so much information to choose from.  Now when an artist tries to copy a human product, say a one hundred dollar bill, he can, with much less difficulty [than he would have in copying a scene in nature], produce something that looks just like the original.  It is easy for him here to hide his personality.  Moreover, he doesn't have to be worried about being limited in realism by the style he inherited from his period and country.  In the case of portrayal (as in portraying a landscape) the correct picture is more like a good map than like a forged bill.  The artist achieves accuracy here by partaking of a long tradition both in his discipline and in his own career.  Various schemata are put forth, and then, based on experience, these are corrected.  The stylistic device of Western perspective, for example, has benefited from such a history.  Some believe that a realistic drawing is a "faithful record of a visual experience."  But it is, rather, a good model that shows all the relations accurately, like a map.  So, even though the artist is influenced by his or her own personal vision and also by the conventions of that tradition, he or she can still make such a model (i.e. such a painting or drawing) to as high a degree of accuracy as he or she wishes.  The question then is what degree of accuracy is required.  Images serve different purposes in different societies and the form of the representation should be understood in terms of its purpose.  Although Gombrich holds to an information-based theory of pictorial realism he does allow for cultural differences, for example two paintings being equally realistic even though they are in different styles:  they just give us different relational models that convey different sets of the total information available.  Moreover, different societies may require different degrees of accuracy depending on their needs.