Eye tracker calibration: How well can humans refixate a target?

Behav Res Methods. 2024 Dec 19;57(1):23. doi: 10.3758/s13428-024-02564-4.

Abstract

Irrespective of the precision, the inaccuracy of a pupil-based eye tracker is about 0.5 . This paper delves into two factors that potentially increase the inaccuracy of the gaze signal, namely, 1) Pupil-size changes and the pupil-size artefact (PSA) and 2) the putative inability of experienced individuals to precisely refixate a visual target. Experiment 1 utilizes a traditional pupil-CR eye tracker, while Experiment 2 employs a retinal eye tracker, the FreezeEye tracker, eliminating the pupil-based estimation. Results reveal that the PSA significantly affects gaze accuracy, introducing up to 0.5 inaccuracies during calibration and validation. Corrections based on the relation between pupil size and apparent gaze shift substantially reduce inaccuracies, underscoring the PSA's influence on eye-tracking quality. Conversely, Experiment 2 demonstrates humans' precise refixation abilities, suggesting that the accuracy of the gaze signal is not limited by human refixation inconsistencies.

Keywords: Accuracy; Calibration; Pupil size artefact; Pupil-CR eye tracker; Retinal eye tracker.

MeSH terms

  • Adult
  • Calibration
  • Eye Movements* / physiology
  • Eye-Tracking Technology*
  • Female
  • Fixation, Ocular / physiology
  • Humans
  • Male
  • Pupil* / physiology
  • Young Adult