Increased Trial-By-Trial Neural Variability Associated with Increased Autistic Traits in Healthy Adults
Objectives: As it is increasingly understood that autistic traits are expressed on a continuum in the general population, the current study expands upon these results to examine whether trial-by-trial variability in neural response is also associated with the occurrence of autistic traits in healthy adults. We utilize a biological motion perception paradigm to explore how response variability is associated with autistic traits under both social and nonsocial contexts.
Methods: Eighty-six healthy adults (49 males) aged 18 to 25 years passively viewed alternating 24-s blocks of point-light-walker displays of biological motion or random motion while undergoing fMRI. Autistic traits were assessed with the Autism Spectrum Quotient Questionnaire (AQ) (Baron-Cohen et al., 2001). An independent components analysis (ICA) was first performed to identify regions of interest (ROIs) with a model-free, data-driven approach. To assess the evoked response to each stimulus type (biological, random), the peristimulus timecourse from each ROI was extracted separately for each trail (5 per stimulus type), with onset aligned to the onset of the stimulus, and a duration of 44 seconds capturing the entire hemodynamic response to the stimulus including recovery. Neural variability was calculated as the standard deviation of each time point within each individual’s timecourse for each ROI and stimulus type.
Results: AQ scores showed a normal distribution and ranged from 4 to 32 in the current sample. The first component of the ICA accounted for 46.57% of variance in data and consisted of 12 spatial clusters encompassing bilateral fusiform gyrus, posterior superior temporal sulcus, precuneous, rostromedial prefrontal cortex, superior frontal gyrus, left orbitofrontal cortex and rostrolateral prefrontal cortex, and right inferior frontal gyrus. Across both conditions and all ROIs, AQ score and neural variability showed a significant positive association (all p’s < 0.0001). For the biological condition, variability of neural response across ROIs accounted for 5 to 28% of variance in autistic traits. Similarly, for the random condition, variability of neural response across ROIs accounted for 7 to 59% of variance in autistic traits.
Conclusions: Previous research has indicated that poor response reliability may be a fundamental neural characteristic of autism. The current study expands upon this hypothesis by demonstrating that healthy adults with a high occurrence of autistic traits show a similar “noisy” neural response across repeated trials of stimulus presentation as indicated by increased standard deviations within the timecourse. Within healthy populations, neural noise has been associated with developmental processes (Misiæ, et al., 2010), and may therefore prove to be a particularly useful metric for informing differential developmental trajectories among individuals with ASD.