The pupil is faster than the corneal reflection (CR): Are video based pupil-CR eye trackers suitable for studying detailed dynamics of eye movements?

• Many video based eye trackers use the p-CR (pupil minus corneal reflection) technique.
• The dynamic properties of the pupil and the CR are different during a saccade.
• The combined p-CR signal overestimates saccade peak velocity and post saccadic oscillation.
• The p-CR signal is not suitable for studying saccade dynamics.

Most modern video eye trackers use the p-CR (pupil minus CR) technique to deal with small relative movements between the eye tracker camera and the eye. We question whether the p-CR technique is appropriate to investigate saccade dynamics. In two experiments we investigated the dynamics of pupil, CR and gaze signals obtained from a standard SMI Hi-Speed eye tracker. We found many differences between the pupil and the CR signals. Differences concern timing of the saccade onset, saccade peak velocity and post-saccadic oscillation (PSO). We also obtained that pupil peak velocities were higher than CR peak velocities. Saccades in the eye trackers’ gaze signal (that is constructed from p-CR) appear to be excessive versions of saccades in the pupil signal. We conclude that the pupil-CR technique is not suitable for studying detailed dynamics of eye movements.