Local and global contributions to direction integration performance in children with autism spectrum disorder

Background: Difficulties in global motion processing have often been reported in autism spectrum disorder (ASD), and have been interpreted as reduced integration of local motion signals in individuals with ASD.  However, these findings have been almost exclusively based on the motion coherence paradigm, which is not a pure measure of integration (Dakin & Frith, 2005).  Indeed, elevated motion coherence thresholds could also arise from imprecision in estimating the direction of individual elements, as predicted by accounts of increased neural noise in ASD (Simmons et al., 2009). The equivalent noise paradigm (e.g., Dakin, Mareschal & Bex, 2005) allows the relative contributions of local and global processes to be separated, by providing independent estimates of local noise (i.e., imprecision in estimating the direction of individual elements) and global pooling (i.e., efficiency in averaging across local estimates).

Objectives: We sought to investigate whether elevated motion coherence thresholds in children with ASD can be attributed to atypical levels of local noise and/or atypical global pooling.

Methods: We presented an equivalent noise direction discrimination task and a motion coherence task in slow (1.5 deg/sec) and fast (6 deg/sec) speed conditions to children with ASD aged 6 to 13 years (n=23) and age- and ability-matched typically developing children (n=23).  In both tasks, stimuli were unlimited lifetime random dot patterns presented for 400ms. In the equivalent noise task, two external noise conditions were interleaved. In the ‘no noise’ condition, the standard deviation of dot directions was 0 degrees, and the threshold was taken as the finest direction discrimination possible. In the ‘high noise’ condition, the mean dot direction was fixed at 45 degrees, and the maximum amount of tolerable noise was assessed. The thresholds from these two conditions were used to fit an equivalent noise function and estimates of local noise and global pooling were obtained. In the motion coherence task, a proportion of dots moved coherently leftwards or rightwards, and participants were asked to indicate the direction of coherent motion.

Results: Children with ASD had comparable motion coherence thresholds to typically developing children.  Additionally, children with ASD had similar levels of local noise as typically developing children. Unexpectedly, however, children with ASD were able to globally average over more estimates than their typically developing peers. 

Conclusions: Three important implications emerge from our results.  First, our findings challenge the common assumption that children with ASD have difficulties with global motion integration and instead suggest that children with ASD integrate more efficiently than typically developing children.  Second, our results suggest that previously reported difficulties in motion coherence tasks may be due to difficulties segregating signal from noise in individuals with ASD rather than reduced global integration. The fact that we did not find elevated motion coherence thresholds in the current study suggests that such difficulties may be dependent on specific task and stimulus parameters. Third, our results challenge accounts of increased neural noise in ASD, as children with ASD had similar estimates of local noise and tolerated more external noise compared to typically developing children.