Impaired Oculomotor Behavior of Children with Developmental Dyslexia in Antisaccades and Predictive Saccades Tasks

Abstract

Analysis of eye movement patterns during tracking tasks represents a potential way to identify differences in the cognitive processing and motor mechanisms underlying reading in dyslexic children before the occurrence of school failure. The current study aimed to evaluate the pattern of eye movements in antisaccades, predictive saccades and visually guided saccades in typical readers and readers with developmental dyslexia. The study included 30 children (age M = 11; SD = 1.67), 15 diagnosed with developmental dyslexia (DG) and 15 regular readers (CG), matched by age, gender and school grade. Cognitive assessment was performed prior to the eye-tracking task during which both eyes were registered using the Tobii® 1750 eye-tracking device. The results demonstrated a lower correct antisaccades rate in dyslexic children compared to the controls (p < 0.001, DG = 25%, CC = 37%). Dyslexic children also made fewer saccades in predictive latency (p < 0.001, DG = 34%, CG = 46%, predictive latency within -300-120 ms with target as 0 point). No between-group difference was found for visually guided saccades. In this task, both groups showed shorter latency for right-side targets. The results indicated altered oculomotor behavior in dyslexic children, which has been reported in previous studies. We extend these findings by demonstrating impaired implicit learning of target's time/position patterns in dyslexic children.

Keywords: antisaccades; dyslexia; eye movements; implicit learning; predictive saccades.

Figure 1

The schematic picture of the task's design. In task (A), the predictive saccades task had constant time and stimuli positions. In task (B), the antisaccades task required an eye movement in the opposite direction of the stimuli, which could appear in one of 3 different positions. In task (C), the visually guided task saccades were triggered in the direction of the target, which could appear in one of 3 different positions.

Figure 2

The Antisaccades task with the percentage of correct antisaccades (AS), error pro-saccades (PS, saccades in the direction of the stimuli) and corrected error (pro-saccade followed by correct antisaccade). Statistically significant differences were found between groups for correct antisaccades (p < 0.000) and error pro-saccades (p < 0.05). The error bars represent standard error.

Figure 3

The Predictive saccades task. The percentage of saccades plotted for the groups according to the latency as predictive (−300–120 ms), regular (120–300 ms), anticipated (< −300 ms) and late (>300 ms). Statistically significant differences were found between groups for predictive (p < 0.001) and regular saccades (p < 0.05). The error bars represent standard error.

Figure 4

Saccadic latencies for each target in the Predictive task. The mean saccadic latencies within predictive and regular type were averaged for every target along the block. The data were smooth with the Cubic smooth splines and the confidence interval was estimated using 10,000 bootstrap replicates (error bar).