Maintaining fixation by children in a virtual reality version of pupil perimetry

Abstract

The assessment of the visual field in young children continues to be a challenge. Children often do not sit still, fail to fixate stimuli for longer durations, and have limited verbal capacity to report visibility. Therefore, we introduced a head-mounted VR display with gazecontingent flicker pupil perimetry (VRgcFPP). We presented large flickering patches at different eccentricities and angles in the periphery to evoke pupillary oscillations, and three fixation stimulus conditions to determine best practices for optimal fixation and pupil response quality. A total of twenty children (3-11y) passively fixated a dot, counted the repeated appearance of an animated character (counting task), and watched an animated movie in separate trials of 80s each (20 patch locations, 4s per location). The results showed that gaze precision and accuracy did not differ significantly across the fixation conditions but pupil amplitudes were strongest for the dot and count task. The VR set-up appears to be an ideal apparatus for children to allow free range of movement, an engaging visual task, and reliable eye measurements. We recommend the use of the fixation counting task for pupil perimetry because children enjoyed it the most and it achieved strongest pupil responses.

Keywords: Eye movement; attention; eye tracking; pupillometry; saccades; virtual reality.

Figure 1.

The three fixation target conditions used in this study; a red fixation point (A), an animation video (B), and a counting task (C). Children were seated in a chair where the headset was positioned. A picture of a participant at home (6 years old) is shown in (D). All fixation targets were displayed at a fixed position in the middle of a dark blue virtual reality environment. The 2 Hz flickering yellow-and black stimuli consecutively appeared at the 20 stimulus locations (E). To ensure accurate retinotopic stimulation stimuli were presented in a gaze-contingent manner (F), i.e., online correction of stimulus locations for saccades from fixation target. Note that thin white lines were added to the background to create a sense of depth in the virtual reality environment. Note that the green gaze position cross was not shown during the experiment but is here shown to illustrate the gaze-contingent presentation paradigm.

Figure 2.

Relative pupil diameter over time for all subjects across fixation target conditions are shown in (A). Pupil traces are averaged across stimulus locations and participants. A raincloud plot depicting the average pupil oscillation powers per fixation target condition are plotted in (B) where the video fixation task (left) is red, the counting task (middle) is blue, and the fixation dot task (right) is green. Individual participants and their age (in years) and test location (i.e., lab or at home (n = 4)) are plotted across fixation conditions. The results show no distinct differences in patterns across age groups or test locations. The fixation dot and counting task conditions provided significantly larger pupil powers than the video fixation target (* = p < .05, ** = p < .01).

Figure 3.

Gaze accuracy and precision of participants; accuracy is defined as the mean gaze deviation from fixation target and precision as the standard deviation of this gaze deviation (A). The gaze accuracy (B) and precision (C) did not significantly differ across fixation target conditions.