Saccade adaptation deficits in developmental dyslexia suggest disruption of cerebellar-dependent learning

Eye movements in dyslexia

Language, writing, and reading deficits are principle elements in identifying children with developmental dyslexia. However, a much broader range of behavioral differences can be observed. Motor learning deficits, discoordination, imbalance, and reduced multisensory and motor processing speeds are often observed [1–7], and several studies of eye movements have revealed differences in saccadic control in dyslexia [8–13]. For example, a large population study demonstrated reduced performance in an anti-saccade task [8], and in a study that employed a double-step eye movement task, increased latencies of orthogonal (but not co-linear) second saccades were observed in dyslexic participants [13]. While these studies provide clues about neural mechanisms that may be altered in dyslexia, neither anti-saccade deficits nor problems in the context of a double-step task are proximal causes of reading disturbances in children with dyslexia. On the other hand, fine control of saccade accuracy and precision is clearly of critical importance to fluent reading [14]. In children with dyslexia, if saccades across text do not land at the expected locations, text might well appear to shift relative to the reader rather than remaining stationary as it will when eye movements land reliably at the commanded positions. Similarly, imprecise eye movements might require multiple corrective saccades to progress along a line of text or to move easily from the end of one line to the beginning of the next. Disruption of these movements will prolong reading times and make comprehension more difficult. It is instructive that reading without needing to make saccades can increase reading rates by 30–40% without degrading comprehension [15], suggesting that even in typically developing children saccades may limit reading rates.

Saccadic adaptation and the cerebellum

When saccades become systematically inaccurate due to eye muscle weakness or, in the lab, by artificially shifting the visual input using prism goggles, the cerebellum and brainstem circuitry operate to reduce the resultant errors [16–19]. Visual errors can also be experimentally introduced during or just after a saccadic eye movement [16, 20–31]. When such errors are persistent, the cerebellar circuitry operates to alter the amplitude of saccades such that the resultant errors are reduced [20, 32–34]. Thus, visual errors trigger cerebellar and brainstem control circuits to rapidly adjust the movement commands in order to maintain saccade accuracy. Short-term adaptation tasks in which a visual error is surreptitiously introduced provide a systematic way to evaluate the function of the circuits involved in the neural control of visual orienting. A failure of the cerebellum in this regard can be revealed in the inability to alter saccade amplitudes leaving saccades inaccurate [18, 35–41].

Cerebellum and dyslexia

The data on structural differences in the cerebella of people diagnosed with dyslexia are somewhat variable ([42–46]). For instance, Brambati [43] found consistent reductions in the volume of gray matter in the deep cerebellar nuclei, whereas Brown et al. [47] reported reduced gray matter in the right hemisphere of lobe VII of dyslexics compared to the left hemisphere (in right-handed participants). Rae et al. [48] also reported a distinct left-right difference in cerebellar hemispheres in dyslexics. They found that in control participants (“good readers”), there was a clear right > left asymmetry in the cerebellar hemispheres but that this laterality difference was not seen in their dyslexic participants. Post-mortem studies [49] report fewer small cells and more large cells in the medio-posterior cerebellum (lobules VI and VII), and Eckert and colleagues [50] could correctly characterize 72% of dyslexics and 88% of controls based on a cerebellar structural model.

Neural activity in lobes VI and VII of the cerebellum has also been studied in people with dyslexia. Nicolson et al. [53] showed reduced right lobe VI activity during an implicit motor learning task in adults with dyslexia. This group has also reported anatomical differences in the olivo-cerebellar pathway in adults with dyslexia [49]. Menghini et al. [51] reported altered lobe VI activity during motor learning, but in this case, the activity was higher in dyslexia compared to controls. This region of the cerebellum is critical for maintenance of saccade accuracy (endpoint error) and precision (endpoint variability). Further evidence of disordered cerebellar activity in dyslexia is provided by the eyeblink conditioning experiment of Nicolson and colleagues [52]. Based on these and other data, Nicolson and Fawcett [7, 53] proposed that cerebellar dysfunction that exists from birth may be a root cause of the symptoms of dyslexia. One prediction of this cerebellar-dyslexia hypothesis is that behaviors which depend crucially on the proper functioning of the cerebellum will be reduced or missing in people with dyslexia. To test this, we investigated saccadic adaptation in people with dyslexia to determine if, in response to persistent visual error, they are able to alter the amplitude of saccades. As outlined above, saccadic adaptation is well known to depend on an intact cerebellum. In this study, as in many previous adaptation studies, a visual error is introduced by moving the target after a saccade has been initiated. This saccadic adaptation task [20] reliably leads to systematic changes in saccade amplitudes in typically developing humans as well as non-human primates [16, 22, 27, 29, 54] and depends on the proper function of the posterior cerebellar vermis [18, 37, 55, 56]. It may function as a new and powerful tool for identifying dyslexia that arises from a disordered cerebellum.

Participants

Tasks

Visual targets were presented on a computer monitor placed in front of each participant. Initial control trials consisted of presentation of a central target (the “T0” target) that the participants were required to fixate within ± 2° for between 500 and 1500 ms. At the end of this interval, the T0 target was turned off and a new target (T1) was illuminated 12° to either the left or right of the T0 location. Participants were instructed to look at the new target. Targets in other locations were also used in this pre-adaptation epoch. After 40 control trials, adaptation trials began. Adaptation trials were similar to control trials, but as the participant made a saccade to the T1 location, the target was turned off and relocated so that it was only 9° (T2) from the original fixation location (T0). If the participant made an accurate saccade to the T1 location, there would appear to be a 3° overshoot since the target was now at the T2 location. The repeated, surreptitious introduction of a visual error at the end of a saccade can drive sensorimotor adaptation [20].

Analysis

Eye movements were measured using an EyeLink1000 (SR Research, Ottawa, Canada). Saccades were identified and movement amplitudes (change in position from start to end of marked movements) stored for off-line analysis. Amplitudes of primary saccades (initial movements from T0 to the T1 target location) were measured and plotted as function of the order in which they occurred during the experimental session (trial number). Lines of best fit (least squares) and 95% confidence intervals were calculated. Each participant was categorized as having adapted or having failed to adapt based on the statistical significance of the slope of the line of best fit; no adaptation occurred when the 95% confidence interval of the slope included zero. Ratios of adapters to non-adapters for the TD and DYS groups were then calculated and a χ ² test was used to determine whether these ratios differed. In addition, the mean amplitude of the first 10 adaptation trials was compared to the mean amplitude of the last 10 using a t test (single tailed, p = 0.05). All analyses and statistical tests were accomplished using Matlab (MathWorks, Natick, MA).