
The following is an expanded excerpt from my February 4, 2009
lecture at the Brooklyn Public Library. You can find additional excerpts under the
theory label.
Getting back to our relativistic thermometer from
Part 1, an analogous situation occurs with our vision. Let’s take a look at some carefully controlled situations below. When presented with a graduated gray scale, the mind receives the available information, organizing it according to the extremes. The brightest field is interpreted as white, while the darkest—the background—looks black.

However, this proves false when seen in a wider context. In the next image, I've added a "true" white square to the top of the scale, shifting our former brightest value down a notch. But had I not done this, our previous set of values would have been just fine for a fully-rendered painting. This is one of the main reasons why it's a good idea to begin with the middle tones. By saving your most extreme "notes" until the end, your decisions will be all the more apt because they were made in context. Alternatively, you can do a preliminary color study so that you can paint more directly on the final work. The example that I most often use is with portraiture: if you want the highlight on your subject's nose or eye to really gleam, every other color must be lower in value, even if it's not pure white (again, your brightest option).

I should note that this demonstration is really only effective in slide form, where the environment is dark, as opposed to here, where the blog background is white.
This principle has serious ramifications when it comes to color (although we are still mostly concerned with value at this point). It means that any and all color is subject to environment: an orange dot on a gray background is construed differently depending on the surroundings. In the context of an illustration, the same color stimuli (physically identical light waves) are interpreted as both an orange painted dot on a black square, as well as a glowing orange dot on a white square—same colors, opposite interpretations.

I apologize for not having the credit information for the image on the left—I found it on-line some time ago. The image on the right was created by extending the gray squares to encompass the entire background, thus revealing the dots' true relationship. Update: here's the original source of the image.
This is a well-established phenomenon, but what does it mean for representational artists? First of all, it means that we have jobs. If this were not true, then paintings would be a very poor description of the surrounding world because the artist’s palette is so small a portion of the physical light gamut. We can’t even look at the sun without damaging our eyes, yet a daub of lead white can faithfully represent it (if placed in the right context). Second of all, we (artists) shouldn’t think of colors the way they are generally taught to us. The names we assign to them can actually be a burden because it associates those colors with certain phenomena, implying an absolutism that prevents us from seeing the reality and, therefore, reproducing it. However, paint is more or less absolute, so the trick lies in your orchestration of the available options. By formulating—and sticking to—a heirarchy of value, the mud you slap on a flat surface can create the illusion of space.