Due to the large expanse of white around the circles, once again our peripheral vision is interpolated by the brain. The black dot is rigidly defined, whereas everything else fades away to white. After a while, the brain forgets that a series of shaded circles exist, and fills in the information gaps with pure white.
In the same diagram on the right, it is impossible to achieve this, as no white surround is present. Also, the outside edge of the circle itself is now precisely defined by the edge with the background.
The same effect is possible with colours, but black and white shades work best. I wonder why this might be? What are these rods and cones anyway?