[W]hat you do when you start to do a painting is that you begin with a basic idea, a hypothesis of what you're setting out to do (a figurative painting or nonfigurative or whatever). Say you're going to paint a figurative painting that's going to be about that model over there and the trees outside behind her and the oranges on the table. It's just a million yet-no decisions. You try something in the painting, you look at it, and you say "N-n-no." You sort of erase it out, and you move it around a little bit, put in a new line; you go through a million weighings. It's the same thing, the only difference is the character of the product.
Let's say at a particular point the scientist gets what he set out to get, he arrives at what he projected might happen if he mixed the particular right combination of chemicals in the right way. But the same is true of the artist: when he finally gets the right combination, he stops, he knows he's finished.