The Cerebellum

, Volume 10, Issue 1, pp 70–80

Greater Disruption to Control of Voluntary Saccades in Autistic Disorder than Asperger’s Disorder: Evidence for Greater Cerebellar Involvement in Autism?


    • Centre for Developmental Psychiatry and Psychology, School of Psychology and PsychiatryMonash University
  • Nicole Rinehart
    • Centre for Developmental Psychiatry and Psychology, School of Psychology and PsychiatryMonash University
  • Bruce Tonge
    • Centre for Developmental Psychiatry and Psychology, School of Psychology and PsychiatryMonash University
  • Owen White
    • Department of NeurologyRoyal Melbourne Hospital
  • Joanne Fielding
    • Centre for Developmental Psychiatry and Psychology, School of Psychology and PsychiatryMonash University
    • Centre for NeuroscienceUniversity of Melbourne
    • Department of NeurologyRoyal Melbourne Hospital

DOI: 10.1007/s12311-010-0229-y

Cite this article as:
Stanley-Cary, C., Rinehart, N., Tonge, B. et al. Cerebellum (2011) 10: 70. doi:10.1007/s12311-010-0229-y


It remains unclear whether autism and Asperger’s disorder (AD) exist on a symptom continuum or are separate disorders with discrete neurobiological underpinnings. In addition to impairments in communication and social cognition, motor deficits constitute a significant clinical feature in both disorders. It has been suggested that motor deficits and in particular the integrity of cerebellar modulation of movement may differentiate these disorders. We used a simple volitional saccade task to comprehensively profile the integrity of voluntary ocular motor behaviour in individuals with high functioning autism (HFA) or AD, and included measures sensitive to cerebellar dysfunction. We tested three groups of age-matched young males with normal intelligence (full scale, verbal, and performance IQ estimates >70) aged between 11 and 19 years; nine with AD, eight with HFA, and ten normally developing males as the comparison group. Overall, the metrics and dynamics of the voluntary saccades produced in this task were preserved in the AD group. In contrast, the HFA group demonstrated relatively preserved mean measures of ocular motricity with cerebellar-like deficits demonstrated in increased variability on measures of response time, final eye position, and movement dynamics. These deficits were considered to be consistent with reduced cerebellar online adaptation of movement. The results support the notion that the integrity of cerebellar modulation of movement may be different in AD and HFA, suggesting potentially differential neurobiological substrates may underpin these complex disorders.


AutismAsperger disorderCerebellumVoluntary saccadeOcular motor

Pervasive developmental disorders (PDDs) are arguably the most serious child onset neurodevelopmental conditions, with a combined prevalence of one in 160 Australian children aged between 6 and 12 years, and unknown aetiology [1]. The main forms of PDD are autistic disorder (autism) and Asperger’s disorder (AD). At present, these disorders are dissociated in the Diagnostic and Statistical Manual of Mental disorders, 4th edition (DSM-IV-TR) [2]; however, the clinical overlap between them continues to fuel debate as to whether they are variants of a single disorder, or whether each has a distinct neurobiology and aetiology. Clinical and diagnostic features common to both autism and AD include deficits in social cognition and communication, restricted interests and repetitive behaviour. In the presence of normal intelligence, high functioning autism (HFA) is diagnosed where language acquisition is delayed [2]. Although not a diagnostic criteria, motor abnormalities are feature of both disorders, and are now considered to be an important early sign of autism spectrum disorders and may be important for early identification [35]. A greater understanding of the nature of motor deficits in these disorders may inform current theories of the neurobiological basis of autism spectrum disorders and the “separateness” of the HFA and AD.

There have been very few studies that compare motor deficits in autism and AD. It has been suggested that cerebellar motor deficits may be more pronounced in autism than AD, while fronto-striatal motor deficits may be common to both disorders [69]. Cerebellar-like deficits have been observed in two studies of gait in young people with HFA; specifically, greater difficulty walking along a straight line, and the coexistence of variable stride length and duration. Conversely, no quantitative gait deficits were observed in young people with AD group relative to control participants [8, 9]. Furthermore, in a study of upper body kinematics [7], individuals with HFA, but not AD, also had longer deceleration phases of movement, in which the final adjustments to movement are made to ensure target acquisition. This is consistent with difficulties seen in cerebellar patients who have difficulty modulating movement to accurately gain targets [10].

The structural basis for altered cerebellar modulation of movement has been well documented in autism, with cerebellar abnormalities among the most consistent neuroanatomical findings [1113]. These abnormalities have included smaller cerebellar volume [14], reduced size of vermal lobules VI and VII (ocular motor vermis) [15, 16], and reduced number [17] and size of Purkinje cells [18]. Recently, reduced connectivity of cerebral and cerebellar motor systems has been proposed in autism, in addition to reduced cerebellar activation during a simple motor task [19]. Cerebellar abnormalities have also been reported in AD, although changes appear less pronounced. While there is some evidence of structural changes, including grey matter deficits [20] and reduced cerebellar volume [14], this has not been consistently demonstrated [20]. It has also been proposed that cerebellar-cerebral feedback projections may be altered in AD [21]. The neurobiological basis for fronto-striatal deficits in these disorders has also been demonstrated [22, 23]. Recently, regional structural abnormalities of the basal ganglia were demonstrated to correlate with the motor, social and communication deficits in autism spectrum disorders (ASDs; this is the term often given to groups of individuals with diagnoses including autism, AD, and PDD) [24]. Characterising, and potentially dissociating, cerebellar motor deficits in HFA and AD is difficult given the interconnected nature of neural motor systems [25]. Nonetheless, further understanding of cerebellar motor deficits in these disorders may be achieved through the profiling of preserved and abnormal ocular motor behaviour since the study of eye movements permits sensitive, non-invasive investigation of motor performance with a high degree of measurement accuracy [26].

A range of ocular motor investigations have been undertaken to explore the behavioural manifestation of the cerebellar neuroanatomical abnormalities in autism and Asperger’s disorder. Importantly, the cerebellum is a key component of the ocular motor system, which incorporates a host of subcortical and cortical structures. As such, it plays a pivotal role in controlling eye movements. There remains, however, a lack of consensus on the integrity of the cerebellar control of ocular motor function in these individuals. For example, deficits in smooth pursuit performance have been postulated to reflect both cerebellar and fronto-striatal dysfunction [27]. In contrast, the demonstration of a preserved VOR and fixation have led to the proposal that cerebellar-brainstem networks [28] and lobules IX and X [29] are spared in HFA. Other, apparently opposing views regarding cerebellar integrity in ASDs emerge from studies of reflexive saccades, with differences seemingly attributable to the way in which performance is measured. Where mean measures of performance (latency, accuracy, duration, and peak velocity) have been found comparable to control participants, it was proposed that cerebellar function was intact [3032], (although see Luna et al. [30], for discussion of possible transitory cerebellar driven impairment in accuracy in young HFA children). In contrast, where variability in saccade accuracy (even with preserved mean accuracy) was found to be increased in two HFA groups (with, and without language delay), it was considered that the cerebellum, specifically the cerebellar vermis or its output through the fastigial nuclei may be functionally impaired in these groups [33]. Furthermore, based on the differential pattern of deficits relative to the TD group, the authors proposed that the two HFA groups “may be fundamentally different at the level of the cerebellum” ([33] p. 1361).

Researchers investigating control of volitional saccades have thus far not commented on the integrity of the cerebellar modulation in ASDs. Volitional saccades have been investigated in groups of individuals with HFA, AD, or mixed ASD groups, primarily using memory guided (ocular motor delayed response) and anti-saccade tasks, with a focus on executive functions as well as motor control. These tasks depend upon higher order executive functions for successful completion, including inhibition of automatic responses, which can alter response times [34] and spatial working memory or spatial transformations whereby the accuracy of the saccade depends upon the internal representation of position [35, 36]. Deficits in executive functions have been well documented in ASD populations [3741]. Correspondingly, both HFA and AD groups demonstrated deficits on the volitional saccades tasks that were attributed to higher level functions. The most striking of these findings was consistently increased response suppression errors in ASD, with variable findings of altered response times and accuracy [3032, 4244].

The neurophysiological loci of dysfunction underpinning these deficits has been postulated to include neocortical circuitry, frontal eye fields (FEF), dorsolateral prefrontal cortex (DLPFC), anterior and posterior cingulate cortices, and fronto-striatal circuitry [3032, 4244]. Since the neural circuitry controlling volitional saccades includes projections to and from the cerebellum it is possible that cerebellar motor deficits may have been evident, although unexplored in these previous studies. A simple volitional ocular motor task with minimal inhibitory or spatial memory/transformation requirements may reveal more fundamental cerebellar motor deficits in the initiation and execution of voluntary eye movements, similar to those demonstrated in Takarae and colleagues’ [33] visually guided saccade task. Furthermore, the focus on mean measures of ocular motor function in previous studies of volitional saccades may have masked deficits in the consistency of motor preparation and execution, known to be modulated by cerebellum.

The ocular motor behavioural manifestation of cerebellar pathology has been extensively studied in primates and human subjects. For saccades, the cerebellum, in particular the dorsal vermis and underlying fastigial nucleus, plays a modulating role, influencing accuracy, consistency, dynamics and adaptation through projections to the brainstem. This modulation directly affects every aspect of the online control of saccades and the acquisition of adaptive ocular motor behaviour [4548] (for a review, see Leigh and Zee [26]). As Barash and colleagues [47] propose “The cerebellar cortex is constantly recalibrating the saccadic system… [including] compensating for rapid biomechanical changes such as might be caused by muscle fatigue” (p. 10931). Thus cerebellar dysfunction may manifest as inaccurate, inconsistent saccades with variable endpoints, acceleration profiles and poor online adaptation. Accordingly, individuals with Friedrich ataxia, a genetically inherited ataxia characterised largely by cerebellar-brainstem motor-circuitry deficits, demonstrate altered latency, accuracy, velocity and velocity waveform relationships relative to controls [49, 50].

In the current study, we sought to further characterise cerebellar deficits in rigorously diagnosed groups of individuals with HFA (IQ > 80) and AD, using a simple volitional saccade task adapted from the wider ocular motor literature [51, 52], with minimal demands on executive processes. We included measures known to be sensitive to cerebellar pathology including velocity waveform measures and the variability of ocular motor metrics. Given the neurodevelopmental nature of cerebellar pathology in HFA and AD we expected to see subtle saccade deficits consistent with those observed in the post-acute stages of cerebellar ablation. We anticipated greater cerebellar-like deficits in the autism group compared with the AD group, consistent with the more circumscribed cerebellar neuroanatomical abnormalities in AD and Rinehart and colleagues findings of greater cerebellar-like motor deficits in HFA relative to AD [69].

Materials and Methods


This study was approved by the Human Research Ethics Committees of Southern Health, North East Mental Health, Monash University, and The Catholic Education Office, all located in Melbourne, Australia. Parents of participants provided informed consent prior to the commencement of the study, in accordance with the Helsinki declaration. Written assent was provided by the participants.

Participants with AD (n = 9) or HFA (n = 8) aged between 11 and 19 years were recruited from The Monash Autism Program in the same as way reported by Enticott et al., [53]. Participants were included in the study if they met the relevant criteria of the DSM-IV-TR for autistic disorder or Asperger’s disorder [2], were aged between 11 and 19 years, had normal vision, and did not have any medical, genetic, or current co-morbid neurological or psychiatric conditions. Four experienced clinicians were involved in the differential diagnosis of the recruited participants. Diagnostic information was gathered using the revised Autism Diagnostic Interview [54], structured parent interview, direct child observations, and information from other sources including allied health professionals and teachers. Inter-rater reliability (for two clinicians B.T. and another experienced team member) was previously calculated on a sample of 107 cases of autism and Asperger’s disorder, generated a Cohen kappa of 0.95 for autism and 0.94 for Asperger’s disorder indicating strong inter-rater reliability for those patients [55]. In the current study one AD participant was medicated with clonidine and fluoxetine but withheld medication on the day of the ocular motor testing.

Typically developing (TD) boys (n = 10) were recruited from a local school and the broader community. No TD participant had any neurological, psychiatric or psychological diagnosis, history of acquired brain injury or first degree relative with a PDD. No female AD or HFA individuals responded to recruitment letters for the study so only male TD participants were included. Verbal, performance and full-scale intellectual ability estimates were obtained using the Wechsler Abbreviated Scale of Intelligence or the Wechsler Intelligence Scale for Children, Fourth Edition for those clinical participants who required a comprehensive cognitive assessment for educational planning (n = 6). All participants demonstrated normal ability (IQ > 70) on verbal, performance and full-scale intelligence measures. Since intellectual abilities were assessed on different, although comparable, Wechsler tools, parametric statistical comparison was not considered appropriate. Verbal and performance IQ estimates were therefore classified as low-average (<90), average (90–110), or high-average (>110), and the proportion of participants in each category was compared across groups using Chi-square analysis. There were no significant differences between Groups on Verbal IQ (χ2 = 5.10; df = 4, ns) or performance IQ (χ2 = 7.90; df = 4, ns) classifications. Group characteristics are presented in Table 1.
Table 1

Participant characteristics












15 years and 0 months

1 years and 7 months

14 years and 4 months

2 years and 1 months

14 years and 9 months

1 year and 10 months






















FSIQ full-scale intelligence quotient (IQ), VIQ verbal IQ, PIQ performance IQ


Horizontal displacement of the right eye was recorded using an IRIS infrared tracking system (Skalar Medical, BV, Delft, The Netherlands), with output sampled at 1 kHz. Screen based stimuli were generated using E-Prime Software and displayed on a 22 in. CRT monitor with a screen refresh rate of 100 Hz. Output from the eye-tracker was displayed alongside a control signal generated by E-Prime indicating stimulus change, and a photodiode was placed directly over a non-visible portion of the screen to concurrently record stimulus change in real time. Eye movements were analysed off line using a MATLAB program developed in our laboratory. Velocity data were smoothed using a 20-point, non-causal (no phase error) smoothing average to reduce noise. All threshold velocities were manually checked against eye movement traces to determine reliability of saccade onset and offset, and any irregularities were manually adjusted.

Participants were seated in a semi-darkened room on a height-adjustable chair with their heads stabilised 840 mm from the centre of the screen using a bite bar and chin rest. Bite bars comprised a mouth mould (Unident single impression trays with two-part catalyst-hardened silicone dental putty, Livingstone International Pty Ltd) attached to the chin rest frame.

Volitional Saccade Task

A schematic representation of the volitional saccade task is presented in Fig. 1. The volitional saccade task comprised stimuli presented on a black background. Central fixation and target stimuli were green crosses (26 × 26 mm) subtending 1.75° of visual angle, with a crosshair (10 × 10 mm) at centre. A fixation screen was presented for 1,250 ms at the commencement of each trial, and comprised a central cross flanked by two white boxes, (53 × 53 mm) positioned such that the centre of each box was 10° to either side of fixation. A cue-target screen was subsequently displayed for 2,000 ms, and included targets at the centre of each flanking box and a left or rightward pointing white arrow measuring 23 mm horizontally in place of the central fixation target. The brief presentation (300 ms) of a small, white, centrally located square signalled the end of the trial. Participants were instructed to fixate the central cross and shift their gaze as quickly as possible to the centre of the target cross in the direction of the arrow as soon as the arrow appeared. A total of 64 trials (counterbalanced for direction) were presented in random order in two blocks separated by a break. The task was preceded by an automated calibration task in which targets moved randomly to locations ±5 or 10° from centre, confirmed by a manual calibration task in which participants were required to fixate targets at centre and ±10°. Step-by-step instructions were provided and a practice trial offered.
Fig. 1

Schematic representation of the volitional saccade task. Target crosses were green with a crosshair at centre. A target appeared in the left or right flanking box at 10° of visual angle to the left or right of the centre fixation cross

Data Analysis

Trials excluded from further analysis were those corrupted by (1) blinks, (2) poor fixation/anticipation (including trials in which the subject was not fixated on the central cue at the time of onset of the target cue (deviation of ±1.5° from centre) and anticipatory saccades (onset of eye movement occurring <100 ms after the onset of the cue), (3) erroneous responses (directional errors occurring >100 ms after cue presentation, irrespective of whether a correctional saccade was made), (4) experimenter error (<1% of all trials), and (5) small saccades with amplitude <3°. One of our key variables, the relationship between main sequence parameters of movement (amplitude, duration and velocity of the saccade) is consistent for saccades with amplitudes of 2–3° to 30–36°. For very small saccades duration remains fairly constant and any increase in the amplitude is caused by an increase in the peak velocity [56].

The frequencies of poor fixation/saccadic intrusions during central fixation and erroneous responses (numbered 2 and 3 above) were calculated as a proportion of the total valid trials (trials not corrupted by blinks or equipment error) and were analysed for frequency using the Kruskall Wallis test.

Saccade onset and offset were identified using a velocity threshold of 30°/s. Saccade measures were latency (milliseconds, time between target and saccade onset), duration (milliseconds, time between saccade onset and offset), peak velocity, amplitude of primary saccade and final eye position (FEP; degrees, position differential between saccade onset and offset (primary saccade) or onset and FEP), and accuracy (gain, amplitude of primary saccade/FEP, representing the proximity of the primary saccade to the subjective acquisition of target). Variability in each of these measures was calculated using the coefficient of variation (SD/mean) for each participant. Velocity waveform measures were Q-ratio (peak velocity/mean velocity - in normal populations this is usually 1.6 [26], and skewness of the waveform (acceleration fraction, computed time to peak velocity/total saccade duration). Variability in skewness, peak velocity and amplitude relationships (peak velocity/amplitude), and duration and amplitude relationships (duration/amplitude) which are generally relatively invariant [56, 57] was calculated for each participant using the coefficient of variation (SD/mean) prior to group means analysis. Participant data was averaged for each subject after excluding extreme outliers (>2.3 SD) from participant’s mean.

Mixed model analyses of variance (ANOVAs) with direction as the within subjects variable and group as the between subjects variable revealed no significant interaction between group and direction for any dependent variable, so all data were collapsed across direction for group analyses using a series of one-way ANOVAs (SPSS). Where violations of homogeneity of variance between groups occurred, established using Levene’s test of equality of variance (p<.05), ANOVAs were computed using separate rather than pooled estimates of variability using the Welch Test (this occurred for variability measures in gain, FEP, Q-ratio, peak velocity/amplitude, acceleration fraction). Where the Welch test was used, standard rather than adjusted degrees of freedom are reported to avoid confusion. Post hoc Bonferroni tests or Games–Howell tests (where homogeneity of variance was violated) were used to investigate group differences. In addition to post hoc tests for significance, estimates of effect size using r2 (as a measure of the percentage of variance in the dependent variable that is accounted for by Group) are reported for comparisons between the TD and each clinical group. These statistics were included to assist in the characterisation of qualitative differences versus severity differences between the clinical and TD groups. Error frequency was subjected to group comparison using the Kruskall–Wallis test non-parametric test.


Means and standard deviations for error, latency, accuracy, and waveform measures are presented in Table 2.
Table 2

Group means and standard deviations for ocular motor measures











Errors (%)








Directional errors







Mean measures

Latency (ms)







Duration (ms)







Peak velocity (°/s)







Amplitude primary saccade







Amplitude FEP














Variability in saccade metrics

Latency variability







Duration variability







Primary saccade variability







FEP variability







Velocity waveform

Q-ratio (peak vel./mean vel.)







Acceleration fraction







Variability of acceleration fraction







Variability in peak vel./amp.







Variability in duration/amp.







aSignificantly different from the TD group at p < 0.05

bSignificantly different from the TD group at p < 0.01

cStatistical trend toward difference from AD group


The three groups did not differ on the mean proportion of trials corrupted by unstable fixation and saccadic intrusions, or in the proportion of directional errors.

Measures of Mean Ocular Motricity

ANOVA revealed no significant group differences on measures of mean latency, peak velocity, amplitude of primary saccade, amplitude of FEP, or gain. There was, however, a significant difference between groups in saccade duration F(2, 24) = 6.43, p < 0.01 with post hoc tests revealing significantly prolonged saccade durations in the HFA group relative to the TD group (p < 0.01; r2 = 0.36). The AD group was not statistically significantly different from the TD group (p = 0.081; r2 = 0.345); however, the r2 value was similar to that in the HFA and TD comparison suggesting the prolonged duration was not unique to HFA. The HFA and AD groups were not significantly different.

Variability in Ocular Motor Measures

ANOVA revealed that variability in saccade latency was significantly different between groups F(2, 24) = 4.50, p < 0.05. Post hoc analysis revealed that the HFA group demonstrated significantly greater variability in time taken to initiate a saccade in response to the cue than the TD group (p < 0.05; r2 = 0.31). The AD group did not differ significantly from either TD (r2 = 0.21) or HFA groups. Group differences also emerged in the variability of the FEP F(2, 24) = 5.21, p < 0.05. Games–Howell post hoc tests revealed significantly greater variability in the amplitude of FEP in the HFA group relative to the TD group (p < 0.05; r2 = 0.38). This increased variability was in the context of preserved mean amplitude, see Fig. 2 for an example. The AD group was not significantly different from TD (r2 = 0.13) or HFA groups. There were no significant differences between groups in the variability of the amplitude of the primary saccade or duration.
Fig. 2

Final eye position achieved by the right eye for saccades made toward a target 10° from centre for a TD and an HFA participant. Although an extreme example, this figure demonstrates the greater variability in HFA

Velocity Waveform Measures

There were no significant group differences in the Q-ratio or the acceleration fraction measures between groups However, the variability of the peak velocity–amplitude relationship was significantly different between the groups, F(2, 24) = 4.26; p < 0.05). Post hoc tests revealed significantly greater variability for the HFA group compared with the TD group (p < 0.05; r2 = 0.33) demonstrated in Fig. 3. The AD group was not significantly different from TD (r2 = 0.06) or HFA groups. There were no significant group differences in the variability of duration–amplitude relationships.
Fig. 3

An example of the relationship between peak velocity and amplitude in a TD and b HFA participants, demonstrating the reduced consistency in the normally relative invariant relationship

A trend toward group differences was observed in the variability of the acceleration fraction F(2, 24) = 3.68, p = 0.056. Visual analysis of the means indicated the HFA group had greater mean variability than the AD group. This was confirmed by exploratory statistical analysis t(15) = −2.52; p = 0.037; r2 = 0.28.


In the current study we sought to characterise imputed cerebellar deficits in rigorously diagnosed groups of individuals with HFA and AD, using a simple volitional saccade task, with minimal inhibitory demands. Boys with HFA demonstrated an inconsistent ability to execute motor behaviour, a feature which was not evident in young people with AD. This was demonstrated by relatively preserved mean measures of ocular motricity overall, with significantly increased measures of variability demonstrated on measures of response time, FEP and movement dynamics. This pattern of grossly intact motor kinematics with subtle but significant impairments in motor consistency is strikingly similar to that seen in the gait of young people with autism. Furthermore, the dissociation between autism and AD at the level of motor consistency reported here is also paralleled in the gait data previously published for this population.

The cerebellum is an integral component of the ocular motor network which comprises subcortical structures (basal ganglia, thalamus, superior colliculus) and visual (primary visual and extra-striate) and parietal cortices. Additional (pre)frontal regions are recruited where higher level cognitive control is required in the generation of more volitional saccades (cue interpretation, spatial transformations, or working memory) [58]. The role of the cerebellum in generating saccades is to make them fast, accurate and consistent, both from moment-to-moment and over time [45]. Evidence from primate studies, suggests that variability in saccade metrics may be considered a key feature of impaired cerebellar control of ocular motor function. In experimental lesions studies, ablation of cerebellar vermal lobules VI and VII, with sparing of the deep nuclei in primates has been demonstrated to directly increase variability of response time and variability of accuracy [59]. Furthermore, Barash and colleagues [47] found that trial by trial variability in accuracy persisted in the months post-lesion despite a near complete correction in mean saccade amplitude from initially hypometric saccades immediately post-surgery. Lesions to the fastigial nucleus, to which the vermis projects exclusively, result in slow, inaccurate and variable saccades [47, 60]. Recently, lesions to the cerebellar hemispheres have been shown to delay saccade initiation and increase trial by trial variability in saccade amplitude [61].

Consistent with neurodevelopmental nature of cerebellar abnormality in autism, which includes deficits to lobules V1 and VII [15, 16], we demonstrated subtle ocular motor deficits in keeping with the chronic, rather than acute, cerebellar pathology. We observed increased saccade duration, increased variability in response time and final eye position. In addition, the dynamic relationship between peak velocity and amplitude was more variable. Individual variability in ocular motor behaviour over consecutive trials has not been reported in a trial by trial volitional saccade task, to our knowledge. Previously however, HFA participants have been shown to be more variable in the timing of their movements to a predictable visual cue [42], consistent with our finding of increased variability in response time. While this appears consistent with cerebellar abnormality, it is conceivable that changes to higher level ocular motor control regions, including the FEF and DLPFC, may have contributed to the deficit observed particularly since the task required a decision to initiate a saccade. However, lesions to these regions typically prolong response time (see Leigh and Zee [26]).

The variability in FEP mirrors a recent finding from our research group in which movement execution time and control were investigated in HFA and AD participants using a novel computerised touch screen version of the Fitts aiming task (Papadopoulos et al., personal communication). When required to draw continuous sets of lines between two targets, HFA participants demonstrated normal execution time and peak velocity, but more constant and variable error than the TD group which was attributed to impaired cerebellar prediction and motor preparation. The AD group did not demonstrate this deficit. Finally, the variability in the relationship between peak velocity and amplitude at fixed amplitude of ±10° was significantly greater in the HFA group compared with the TD group, implying a reduced coupling of this normally invariant relationship in HFA. This deficit was also demonstrated in the reduced consistency in the peak velocity and amplitude relationship observed in individuals with Friedreich ataxia [49]. Main sequence relationships are usually explored over a range of amplitudes; however, our participants were examined at ±10° only. It would be important to further explore this finding over a range of amplitudes at which this relationship holds (approximately 3–30°) [56].

Despite the increased variability in a number of measures, the HFA group demonstrated largely age appropriate motor responses when the mean performance was examined. Preserved mean responses were demonstrated for latency, peak velocity, accuracy, and amplitudes of primary saccades and FEP. The exception was prolonged saccade durations relative to TD participants. There has been considerable variability in the previously reported findings on ocular motor behaviour for voluntary saccades in the context higher cognitive demands. Notably, however, peak velocity and duration have been consistently reported as preserved in individuals with HFA [3032]. Our finding of prolonged mean saccade duration the HFA group without corresponding significant changes in peak velocity or skewness of the velocity profile suggests a general slowing in both acceleration and deceleration phases. This differs from the finding in Rinehart’s gait study where the deceleration phase of movement in the HFA group was prolonged. Slowed saccades have been demonstrated with inactivation of the caudal fastigial nucleus [62] and the cerebellar vermal lobules VI and VII where post-lesion, the primates demonstrated increased duration and changes to both acceleration and deceleration, although greater changes were observed in deceleration [59]. Although increased saccade duration is consistent with cerebellar pathology, it is not clear why this has not been demonstrated in HFA before, and may be a result of our small sample size.

The AD group did not differ significantly from the TD group on any measure, although there was the suggestion of prolonged durations similar to those in the HFA group when estimates of effect size were considered. Localised abnormalities demonstrated in the connectivity of the cerebellar outflow pathway [21] may not manifest in abnormalities in the measures we have examined here since feedback to cortex is not necessary for these saccadic movements. Structural cerebellar abnormalities in AD have been demonstrated in altered grey matter concentration [20], however there is as yet little evidence of deficits to cerebellar areas known to directly modulate saccadic eye movements, unlike in autism. There was some suggestion that the HFA group demonstrated greater variability in the skewness of the velocity profile than the AD. This trend toward increased variability in skewness occurred without concurrent increased variability in saccade amplitude, and may reflect the greater cerebellar pathology associated with HFA. This would need to be replicated with larger samples to determine whether this increased variability distinguishes HFA from AD.

In this study, the HFA group demonstrated a range of subtle deficits, particularly in producing consistent movements over repeated trials. That increased variability in motor behaviour should occur across studies of gait, upper limb movements and ocular motor control provides concurring evidence of reduced online modulation of movement by the cerebellum. This may be the result of reduced feedforward online control of movement [63]. The modulation of repeated movements by the cerebellum is thought to depend largely on efference copy error feedback whereby errors from past movements are used to update subsequent movements. Fabbri-Destro and colleagues have recently proposed that young people with autism have a “deficit in chaining motor acts into a global action” [64] p. 521. Our results suggest reduced integration of motor behaviour at the level of the cerebellum, supporting the notion that motor acts are planned independently from one another. The increased variability may reflect the higher degree of independent programming of movement in response to the cue at each trial, rather than a refining of previous motor sequences initiated in prior trials. This reduced ability of the cerebellum to “automatise” the regulation of movement may drive the increased reliance on cortical regions associated with the effortful control of motor execution. This view would be consistent with recent findings of decreased cerebellar activation in HFA on a simple, repetitive motor task [19] and a visually guided saccade task [65].

Another possibility is that the cerebellum is less able to compensate for the subtle biomechanical changes associated with fatigue in these participants [47]. However, this pattern of more variable movements in HFA, but not AD has been demonstrated across studies of gait, upper limb movements and now ocular motor control with tasks ranging in timing from 40 s to a few minutes, thus it appears the decreased consistency in motor production occurs from the outset in this population, likely associated with fundamental neurobiological deficits.

Our finding of reduced consistency in the online control of movement may inform current theories of autism. The cerebellum has been implicated in a range of theories postulated to account for the widespread impairments in autism spectrum disorders [6668]. It is now widely accepted that the role of the cerebellum extends beyond that of motor control, and it has been implicated in cognitive and affective control of behaviour, a role supported by its connections to and from cortex (i.e. cerebro-ponto-cerebello-thalamo-cerebral connections) [6975]. Recently, theories of reduced connectivity of neural systems have gained momentum [19, 67] and neuro-imaging evidence of reduced or altered connectivity providing a neurobiological basis of support for theories of ‘weak central coherence’ [76] and theories of ‘complex information processing’ [77].

This study has revealed some ocular motor findings of significantly altered cerebellar modulation of movement in HFA but not AD boys. A potential weakness of our study was that we chose not to co-vary for any IQ measure since none of the dependent variables correlated significantly with any IQ measure, and Goldberg and colleagues [42] reported that their analyses did not vary with the inclusion of PIQ as a co-variate. A significant limitation to our study was the small sample sizes. We have, however, demonstrated that differences may exist between the HFA and AD group at the level of cerebellar motor control that are observable when variables sensitive to cerebellar pathology are explored. It appears as though, on such measures, there exists both qualitative differences and differences in severity of impairment between the clinical groups. For example, both the HFA and AD groups demonstrated prolonged saccade duration and increased latency variability relative to the TD group, the results indicating that difference between the HFA and AD groups may be a matter of degree of severity; however only the HFA group was significantly different from the TD group. In contrast, the variability in velocity profile measures appears to be a deficit unique to the HFA group. Degree of severity cannot wholly account for the pattern of findings, lending support for qualitative differences in aspects of cerebellar modulation of movement between individuals with AD and HFA. Further ocular motor investigation of both AD and HFA groups with the inclusion of measures of variability would be important to replicate our findings and extend our understanding of the potentially different neurobiological basis of these disorders and the role of the cerebellum in the motor deficits in these groups. Motor investigations may be particularly beneficial to the dissociation of deficits in AD and HFA because once the task is understood, language is not a confounding factor in the comparative measures, which is important to avoid circularity of dissociating HFA and AD based on the findings that relate to diagnostic criteria [78].


On a simple motor task we have demonstrated a number of deficits consistent with abnormal cerebellar modulation of movement in individuals with HFA. The same deficits were not observed in AD participants. While our small sample sizes are a caveat on the extent to which these findings contribute to the discussion of fundamental neurofunctional differences between HFA and AD, our findings are consistent with Rinehart and colleagues gait studies and recent upper limb sequencing findings. Across three motor systems, the HFA, but not the AD group have demonstrated significant deficits in the cerebellar modulation of movement. These findings support the argument that the disorders may have separate, albeit overlapping aetiologies. Categorising them together, as proposed for the fifth edition of the Diagnostic and Statistical Manual [79] may be premature, since symptoms from disorders with discrete neuropathologies may be ameliorated by different treatments.


We gratefully acknowledge the time and effort of all the participants and their families involved in this study. We thank Lynette Millist for technical support. This research was supported by Monash University, an Autism Speaks Award (#CF06-0154) awarded to Dr Joanne Fielding, and two grants from the National Health and Medical Research Council Australia: Fellowship grant (#454811) awarded to Dr Joanne Fielding and Project Grant (#585801) awarded to Dr Joanne Fielding, Dr Nicole Rinehart, Prof B.Tonge and A/Prof O. White.

Conflicts of Interest

There are no conflicts of interest.

Copyright information

© Springer Science+Business Media, LLC 2010