Item

Abstract

Video-based eye-trackers are typically calibrated by instructing participants to fixate a series of dots, the physical locations of which are known to the system. Unfortunately, this procedure does not verify if fixation has actually occurred at the desired locations. This limitation can be remedied by requiring participants to perform a simple visual discrimination task at each location, thus mandating accurate fixation. Still, it remains an open question whether this modification could affect fixation accuracy. In the current study, we compared the accuracy of fixations that were performed with a visual discrimination task and those without such a requirement. Participants either identified the orientation of a small Landolt C (size = 0.1°) or fixated a similar probe without performing the task. Results indicate that participants fixated equally well in both tasks (mean diff. of abs. error = 0.01°, Bayes factor B01 = 4.0 with JZS prior, see [Rouder et al., 2009, Psychonomic Bulletin am
p;am
p;am
p; R
eview, 16(2), 225-237]). Given this, we propose the implementation of this visual discrimination task to eye-tracking calibration protocols as it elicits verifiable fixations without compromising fixation accuracy.