Figure 3A.
Testing for pupil centration artifacts. Left: A pupil response is elicited by making the screen bright for about 50 msec. Right: The measured convergence of both eyes is plotted against pupil size. If pupil centration would change, a linear fit through the data would have a regression different from zero. Here, the regressions were close to zero which mean that no pupil size dependent correction was necessary for the eye tracker.