Abstract
The predominant view of saccade adaptation is that it is largely guided by retinal error–the postsaccadic distance of the target from the fovea. In laboratory experiments with the target as the only visible stimulus, the retinal error is unambiguous, unlike real life, in which any of a multitude of stimuli might be on or near the fovea after any saccade. To take a first slice at this issue with Occam's Razor, we asked whether saccade adaptation would be prevented if a second stimulus, resembling the target, occupied the original target location, while the target was displaced during the saccade.
A target (on each trial randomly either a green circle or a yellow square) stepped by 10-12 deg. On 400 adapting trials, the target made backward (or, in separate experiments, forward) intrasaccadic steps (ISS, 20% of the initial step amplitude). In some experiments the non-target occupied the original target location, while the target made an ISS; in others, the non-target made the ISS.
Surprisingly, the non-target exerted almost no effect on the adaptation. When it was the target that made the ISS, the gain decreased or increased approximately as much whether or not the original target location was occupied by the non-target. Conversely, if the target did not move during the saccade, but the non-target made an ISS, very little adaptation occurred. Finally, if, after adaptation to the target alone, the non-target was added at the original target location, the gain did not return towards normal, but did if it was the target that remained at the original target location.
Conclusions: Although generally considered a low-level motoric phenomenon, saccade adaptation can be selective for the relevant error signal, using the visual attributes of the target from trial to trial for guidance.