Line drawing is a very stylised form of art. There are no lines in nature, and artists have to learn the conventions to be used when drawing in line. I am interested in whether one could help them learn by repeated priming of associations from realistic images to line drawings of those images.
First, here is an example, a drawing of a boy from the children's annual Sunshine Stories, published probably in the 1930s. The top of the chin has been omitted, because it catches the light. The upper lip is shown dark, and the lower lip is omitted, being indicated only by a shadow under it:
This is a very common convention. The link points at an excerpt from an instructional book that was very popular in its time, Simple Sketching in Line by L. A. Doust. For some reason, the visual system seems to pay too much attention to bottom lips that are outlined or dark, so artists have discovered that it's often best only to draw the shadow under them.
Faces are not the only subject that gets conventionalised. For example, at this page and subsequent pages, I show some of the conventions used in indicating texture. These slides, by the way, are from a talk I gave on the semantics of line in drawing, showing the different purposes for which artists use line.
So, imagine that we ask a skilled illustrator to take 500 (say) photos and make a drawing of each, using a common and consistent style. We then use the photos paired with their drawings as priming stimuli. Is there any evidence for or against my hypothesis that this would help novices learn the mapping from realistic images to drawings thereof? In a sense, it's analogous to teaching musicians to sight-read.
A hi-tech alternative to photos and drawings would be mediated-reality glasses. In Oxford, Stephen Hicks is developing Smart Spex, which help people with poor vision to see. The third image on that page shows a scene, labelled "a", and the specs' renderings of it. Rendering "c" is similar to a line drawing. If a subject were to wear these glasses in "line drawing" mode continuously, would they absorb its rendering conventions? If so, it could be a very useful way to teach drawing.