Neural Correlates of Goal-Directed Reaching Movements in Children with Autism Spectrum Disorder

Background:

Children with Autism Spectrum Disorder (ASD) exhibit restricted interests and repetitive behaviors that may represent aversion to environmental uncertainty and an attempt to prevent or minimize change (Gomot and Wicker, 2012).  When environmental change does occur, challenging behaviors can ensue. Thus, brains with ASD appear to respond differently to environmental uncertainty than do typically developing brains.

Objectives:

We used a novel robot and functional magnetic resonance imaging (FMRI) to measure blood oxygenation-dependent (BOLD) signals related to fast, goal-directed wrist movements (reaches) made against predictable and unpredictable robotic loads. We compared movement kinematics and BOLD signal changes produced by high-functioning children with ASD and by typically developing (TD) children. We evaluated two competing hypotheses:  (1) Children with ASD and TD children solve the predictable motion task using similar feedforward control strategies that recruit similar neuronal networks, but that these patterns are fragile in ASD and susceptible to disruption in the presence of environmental uncertainty; (2) children with ASD and TD children solve the sensorimotor task using neural control strategies that differ in recruitment of feedforward and feedback mechanisms.

Methods:

Nine high-functioning children with ASD and eleven TD children reclined in a magnetic resonance imaging (MRI) scanner and held the handle of a plastic robot. They made 250 wrist flexion movements (one per trial) against spring-like loads that either remained constant (trials 1-50) or varied unpredictably across trials (trials 51-250). Subjects were to capture targets on a video screen using a cursor controlled via by the robot handle. We analyzed movement kinematics to obtain a model of each participant's trial-by-trial prediction of upcoming load. The technique quantifies the extent to which sensorimotor memories from past performances influence future movements. We analyzed functional MR images to compare the neural control networks recruited by the two groups of children exposed to the constant and variable loads.

Results:

Both groups used memories of prior performances to adjust aim on subsequent trials. Both groups undershot the target, although children with ASD moved ~10% less than TD children. Movement errors were more variable in ASD. We included the mean and standard deviation of movement error as co-factors in the image analysis to control for possible group-wise bias. Neuroimages from constant load trials revealed widespread BOLD signal activations in the TD group in regions previously implicated in feedforward or feedback control. By contrast, activation volume in the ASD group was ~17% of TD volume and was represented more heavily in the feedback control network; Within select feedback control regions, BOLD impulse responses were markedly larger in ASD vs. TD. For both groups, we found no difference in BOLD signal changes across load conditions.

Conclusions:

Our findings support the hypothesis that children with ASD and TD children solve the target capture task using neural control strategies that differ in their recruitment of feedforward and feedback control mechanisms. The strategy used by the children with ASD appeared to favor feedback control regardless of load condition, even though this strategy resulted in movements that were less accurate and more variable.