What should the tokens be?

gaze direction/target?     gaze transitions?     something else?