A recent comment to another blog post raised some interesting issues so I figured I should bump it up to a full discussion here.
The basic issue is whether drawing in a realistic style (proper anatomy, shading, depth, etc) is somehow antithetical to our cognitive predispositions for drawing. I've argued before that drawing involves the deployment of graphic patterns in our minds in various ways, and that fluency is a proficiency in this system of representation.
There seem to be two issues here regarding the system of drawing:
1) What can we do?
2) What are we predisposed to do?
The answer to the first question is that we absolutely can draw realistically. Indeed, we can even develop mental patterns that can generate these realistic images. This is what creates a "style" of drawing. No one would say that Jim Lee or many other modern comic artists are "cartoony", yet they certainly use consistent patterns in their drawings that can be tied directly to their "styles."
However, I would say regarding question #2 that we are not predisposed to draw realistically. The evidence I've seen seems to suggest that our natural system of drawing is based more on representing contours and basic patterns. The more realistic aspects of a drawing system (perspective, depth, shading) are always the things that are more struggled with and require explicit teaching to learn (i.e. they won't be acquired effortlessly without some form of external instruction).
My advisor has recently been addressing a similar issue in language: what is it that a language speaker naturally is predisposed to know without any exposure to an external system, and what is learned (and relatedly, what aspects of cognition do we share with other primates and what has evolved to uniquely enable humans to speak). His argument, which will hopefully be making an appearance in a journal sometime soon, is that there is a certain set of innate principles that humans have regarding speaking, and that modern language has added structure on top of these deeper predispositions.
This is largely the way I view drawing. Our cognitive predisposition is for certain types of graphic representation: line drawings built of patterns that lack perspective and depth, though can use occlusion, etc.**
On top of this is the potential for explicit instruction for further iconicity: "accurate" anatomy, shading, perspective, etc. This is why much pre-Renaissance art lacked these iconic features. Because we aren't predisposed for it, "point perspective" and "shading" was a "discovery" and not something learners mature into naturally.
This doesn't mean we "can't" or "shouldn't" do it, but it does mean we aren't wired to do it without instruction.
** For an excellent work on what the developmental trajectory of drawings looks like without influence from learning an external system, check out John Willat's Making Sense of Children's Drawings.