Perceptual task drives later fixations and long latency saccades, while early fixations and short latency saccades are more automatic.

We used a simple stimulus, dissociating perceptually relevant information in space, to differentiate between bottom-up and task-driven fixations. Six participants viewed a dynamic scene showing the reaction of an elastic object fixed to the ceiling being hit. In one condition they had to judge the object's stiffness and in the other condition its lightness.