If I understand that picture correctly, the brightness is a function of how many subpixels are turned ON in a given image frame. Your eye integrates all the photons it receives, so if all 16 red subpixels are ON, you'll perceive "red" but also perceive "very bright." If only one is ON, you still perceive "red" but will see "very dark."
edit to add:
It occurs to me that the perceived color could in theory be adjusted by toggling the pixels' states at, e.g., 240Hz to be ON for only part of a video frame. However, this has not been achieved in actual devices, and in any case would defeat one of the desired features of this technology, which is to maintain a color image without drawing additional power (the pixels hold state until addressed).
For those who don't know much about the retina, some info:
There are color cones, typically 3 types, each of which is sensitive to a different band of frequencies. The combination of signals from these cones is what the brain processes to sense a color. Along with these are rods, which are more sensitive detectors but are monochromatic (only report total intensity). Combine the intensity signal with the color signal to get teh final perceived shade.