Introduction: We have previously shown that saccade control measured in an antisaccade task develops continuously in both normally reading and dyslexic subjects until about age 20. Yet, the saccadic performance of dyslexics stays significantly behind that of normally reading subjects at all ages. These results were interpreted as a dysfunction in attention mechanisms) involved in saccade generation as well as in the reading process and probably controlled by the prefrontal cortex. The present study shows how a daily training of specified visual tasks can improve saccade control in dyslexics.

Methods: Seventy-eight dyslexic children and adolescents (age between 8 and 17 y) showing a strong impairment in fixation stability and/or in voluntary saccade control performed one or more of three tasks including a fixation, a saccade, and a distracter (requiring antisaccades) condition. In all tasks the subject had to detect (at fixation or after a saccade) the last orientation of a small pattern which changed its orientation quickly (speed adjustable between 90, 130, 170, 210 ms) between up, down, left and right direction. After some time the pattern disappeared and the subject had to press a key indicating the last direction of the pattern. Subjects practised 10 min a day for a minimum of one week (for one task) and a maximum of 8 weeks (for all three tasks together).

Results: 80% of the subjects showed an improvement of the error rate in a classical anti gap task after training. The average improvement was 19% in the error rate, 13% in the correction rate, and 48 ms in the saccadic reaction times (SRT), which were faster. The data also confirm that the ability of suppressing involuntary glances to the stimulus develops with age. The two curves describing the error rate before and after training at different ages were almost parallel. When compared with the data of 107 control children the mean values of the error rate in dyslexics after training still stayed behind at older ages in spite of the improvement in the performance. Children who trained only the saccade pro task did not show improvements in the anti saccade task.

Conclusions: Reading and writing are complex functions that involve auditory/language as well as visual/attention related brain regions. The finding of a deficit in voluntary saccade control in dyslexic subjects supports the hypothesis of a multifactorial etiology of reading problems including prefrontal functions. This deficit is not definitive, it can be changed by a daily training improving sensorimotor integration.