An attempt to integrate cognitive sciences with art making.

Thursday, September 4, 2008

How To Draw What You See

It is commonplace for beginning artists to be advised to paint "not what they know but what they see. " The problem is understood to be that the novice has a "literal" approach versus the "visual" approach of the advanced artist. The argument goes that we "think", for example, that trees are round green forms, and therefore the beginner paints them the color and shape he thinks them to be: the tree is portrayed as something that resembles a green lollipop. This misperception of the object in spite of visual evidence to the contrary so impressed psychologists Cohen and Bennet in their article "Why Can't Most People Draw What They See?" that they suggested that it was delusional in nature. (footnote 1) What's equally as delusional is the explanation put forth by current day art instruction which I just summarized.

Paradoxically the advice to "draw what you see" is hard to follow. Beginners struggle with it assuming that they are just so accustomed to relying on words, a "literate approach", rather than vision, a "visual approach", that they can't abandon their old habits. It is hard to understand why anyone would believe that the word "tree" should influence the beginning artist to draw a green lolipop. However, like a great deal of theory taught in art school this notion of the literate versus the visual is, at best, based on obsolete19th century ideas. For unknown reasons the art world and the world of scientific studies of vision and cognition have remained isolated from each other. Much could be gained in both fields by more discussion between them.

To begin with, identifying the problem as one of "seeing" or not seeing as the case may be, just isn't accurate. What an artist making representation art is doing is looking at the subject, turning away and looking at the canvas, and then making a mark. This is not about a visual task or for that matter a verbal task. It is a memory task. It involves what is called visual memory as opposed, for example, to verbal memory. Hence my somewhat reversed title: " how to draw what you see" is wrongly put. "How to remember what you see in order to draw it" is more to the point.

But what's so hard about remembering what you see for the minute or two it takes to "attack" the canvas with brush and paint? The hitch is that there are two types of visual memory, short-term visual memory and long-term visual memory, and here's where the contribution of cognitive comes in. It is only in short-term visual memory that an accurate, specific, and particular visual image of the subject is held. In contrast, long-term visual memory stores simplifications and interpretations of images like the lollipop tree that are not appropriate for making realistic representations. Unfortunately short-term visual memory is really short and limited, so short that the image is lost soon after one turns to the canvas and so limited that it can hold only a small bit of the subject. So it is not that art students are substituting what they know, "verbal memory", for visual memory. They are substituting long-term visual memory simplifications and interpretations for short-term visual memory. To repeat the major task of making a likeness is a short term visual memory one.

The importance of grasping this way of looking at the problem of making representational images lies in the fact that it can explain what beginning artists need to learn to do, why and how certain techniques work, which ones seem to be better than others, and, perhaps, generate new exercises, methods, and research.

Short-term visual memory
Short-term visual memory has three drawbacks. It cannot record all the visual information that bombards us, that which it stores it does so for a second or two, and it is not generally "interested" in the minute detail that an artist needs.

It simply would be impossible to store "photographic quality" images of all the real world scenes that fall on the retina any more than this could be done with a camera. It would not take long for both the brain and the camera to run out of storage capacity. In fact, short-term visual memory is quite limited in what it can hold. It cannot hold the entire image that falls on the retina. It holds a surprisingly small area of the scene in front of us. Drawing an eyebrow in one study involved looking at it three times to refresh short term visual memory. So short-term memory's first drawback is the limited amount of real, accurate visual information it can store long enough to be useful to the artist.

Short-term visual memory's second drawback is obvious from its name: that what it does store deteriorates rapidly, in fractions of a second, and has to be refreshed often.

There is a third drawback to short-term visual memory. When we look at a scene we tend to "scan" it whether it's a landscape, in which case we seem to be scanning to make sure there are no predators lurking about, or a person in which case we are trying to obtain from body language or facial expression as much information as we can about them. Just ask any woman about this. The point here is that, to look three times at an eyebrow or anything else in order to depict it in our art work, we have to resist the tendency to look all over the place. This is much harder to do than it seems.

Long-term Visual Memory
It is in long-term visual memory that the image is stored pretty quickly in such a way that it can be held for a long time and retrieved easily. Unfortunately, to do so the image is transformed from a specific representational image of which an artist would want to make a copy into a generic interpretation sometimes called a "schema". A schema resembles its subject in much the same way that a map resembles a landscape and like a map, although it is very useful, it is not a "picture" of the thing mapped. The satellite image does not resemble the map. This transformed image which we remember "seeing" often resembles what children draw.

"Schematic Pressure"
The problem, of course, can be subtler. Schemas can creep into small parts of drawings, and it is a never-ending task to be alert to them. For example, if short-term visual memory is not constantly refreshed and held focused on particular passages, an artist can easily slip and make the iris of an eye round even in three-quarter view or profile.

Fundamental to making representational art is to realize that there is a constant tendency to use the image (schema) we have in long-term memory rather than the more limited and fleeting one in short-term memory. One way to put it is that there is a kind of pressure from schemas to get out and into our art work. If we look at a tree, we can keep in short-term memory only a fragment of that tree for only a brief moment. If we are drawing a limb, we have to constantly return to each new segment of that limb to refresh short-term memory. If we don't, the "schematic pressure" from long-term memory will substitute a simplified and cartoon-like image of tree limbs in place of the real thing.

Representational artists have learned to use short-term or what is sometimes also called more aptly, "visual working" memory. This would at its most simple form require looking many times at small areas of the subject. They must have also learned to not use the long-term memory schemas. But how do they do it and how did they learn it? Answers can be divided into those that are related to improving short-term memory and its rate of refreshing and those that related to decreasing the power and simplicity of long-term memory schemas. The following is not intended as a comprehensive survey, but merely illustrations of how the paradigm suggested above might be applied.

Refreshing short-term visual memory
There is some evidence to suggest that short-term visual memory can be improved by practice. There is an apocryphal story of a French teacher of painting who posed a model on the ground floor and had students set up their easels four stories above it. They had to run the stairs over and over again trying to hold in short-term memory what they wanted to paint. Whether it was better for their cardiovascular conditioning or their art-making is debatable. The opposite, however, is clearly helpful. The less time there is between the look and the putting of a mark down on the artwork the better. John Singer Sergeant's portrait technique would be a case in point. He put his canvas directly beside the subject, somewhat like we do with the sight-size method, and stepped back quite a ways to get visual information in short- term memory. Then he would rush forward to put a mark on the canvas before the short-term visual image deteriorated.

The admonition of one of my teachers that a brush is not an "automatic weapon" and should be used only once before being "reloaded", also seems apt. Here, by restricting oneself to one mark before looking again and refreshing short-term visual memory, it is easier to prevent substituting long-term memory schemas for short-term memory image fragments. "Three looks, two thinks, and one mark" is how it is sometimes said.

It is not only a question of what to do but, also, what not to do.  With enough practice, the artist learns not to move his pencil if he's going beyond what is or was in short term visual memory and to look again.  When to stop the line and look again is the critical moment of insight in drawing representational images.

In drawing a likeness some teachers suggest drawing what might be called a vector diagram. Even if the aspect of the subject to be drawn is a curve, these teachers suggest drawing from point to point and smoothing out the shape later. Helen Van Eyk's reason for suggesting this approach to students was "They look more often" because they break the image down into multiple segments. We now know that means that they refresh short-term memory more, and it is this that recommends this technique over all others.

Contrary to the standard advice thinking "verbally" about what you are looking at seems to help. People with no artistic training tend to scan with a free-floating attention even when they are specifically instructed to make a realistic image. For example, if they are asked to make a portrait of a face, they still do not look at any one form. Studies of which part of the brain is active do not suggest that they are thinking about anything in particular. Paradoxically the visual part of the brain is most active just as the standard advice suggests. On the other hand when a trained artist is asked to make a portrait, he or she tends to zero in on one particular form, and the thinking part of the brain is quite active. This has been demonstrated by having an artist and several non-artists attempt to draw a portrait while simultaneously having a functional MRI of the brain which can determine what parts of the brain are most active.

It is much easier to keep focused on a specific part of a subject if we can think of questions about it, questions that often come from the canon of proportions. If we can look longer, we can hold the image in short-term memory longer. If you try looking at a small part of an image, you will be able to feel the urge to move your eyes to other parts of the form or subject, to fall into the scanning mode. On the other hand, if you are thinking about a question such as "Is the upper lip half as wide as the lower lip?" (From the canon of proportion) you can focus your attention better.
Reducing or Improving Long-Term Visual Memory Schema
Knowledge of the canon of proportions whether it is of the figure or the artistic anatomy of trees will enrich long-term memory schemas. It will change the green lollipop into a more complex cliche but not into an image that looks like the tree before us. Just depending upon the canon of proportions, of course, will produce a realistic-looking "academic" image but not a portrait image. The resulting image will look better, but it will suffer from the same problem of schemas replacing realistic images. It is not useful for the purposes of making a representational image by itself.

If, when painting a portrait, we focus on depicting the shadow shapes, the darks, and focus intensely on them, we are not dealing with anything that is stored in long-term memory as a schema. The shadow on the side of the face over the eye socket and down the side of the nose is a one-of-a-kind form. Because we are not drawing the nose or the eye, but the shadow shapes pressure from schemas is greatly diminished. There may be some pressure to make squares and triangles out of the shadow shapes, but this is much less powerful than the schemas in long term visual memory of the eye and nose. In addition we have to look more frequently because there are no prototypes in long-term memory to fall back on, and there are no canon of proportions for shadows. The portrait emerges in the negative spaces. Emphasis on drawing or painting "the light" rather than the forms has the benefit of bypassing the simplifications of long-term visual memory schemas.

Final thoughts
The artist does not employ short-term visual memory in order to make what a photograph or a plaster life cast would display. It is not an exact portrait or what has come to be called a "photographic likeness" that the artist should be after but a "super portrait". The artist is capable of making an image of a person, tree, or landscape that reads to the brain more like the image than the thing itself. Once one has mastered the "how to look", this is the next challenge. And cognitive science has something to contribute to this as well. I discuss it below in another post.

1 Cohen DJ, Bennet S why can't most people draw what they see? J Exp Psychol Hum Percept Perform. 1997 Jun;23(3):609-2