Dimensional Arousal and Categorical Gaze Instability: Uncoupling the Baseline Oculomotor Phenotypes of Autism and ADHD

BackgroundAutism Spectrum Disorder (ASD) and Attention-Deficit/Hyperactivity Disorder (ADHD) share substantial clinical and physiological overlap. While naturalistic and sensory-driven paradigms increasingly capture evoked neurophysiological responses, the intrinsic baseline physiology of these conditions remains poorly defined. We aimed to characterize resting-state autonomic arousal and oculomotor stability across the ASD-ADHD spectrum using both continuous (RDoC) and categorical (DSM-5) analytical frameworks. MethodsWe analyzed resting-state eye-tracking data from a large pediatric cohort (N = 2,640) from the Healthy Brain Network. During an unconstrained baseline, we extracted Pupil Relative Volatility (Coefficient of Variation [CV]) to index intrinsic autonomic arousal, and the Bivariate Contour Ellipse Area (BCEA) to index spatial gaze instability. Data were evaluated using continuous dimensional regressions against the Social Responsiveness Scale (SRS) and SWAN inventories, followed by 2x2 factorial ANCOVAs based on clinical diagnoses. Sensitivity analyses accounted for clinical collinearity, spatial outliers, and psychostimulant medication. ResultsDimensional models revealed that Pupil CV was significantly and uniquely associated with continuous autistic traits (q = 0.0043, joint model), exhibiting a strong statistical suppression effect when controlling for ADHD trait covariance. However, this pupillary biomarker lost significance in binary categorical models. Conversely, spatial gaze instability (BCEA) demonstrated robust categorical threshold effects; isolated ASD and ADHD diagnoses significantly impaired baseline gaze stability. Furthermore, comorbid ASD+ADHD produced a distinct, sub-additive interaction for BCEA (q = 0.005) that remained robust to extreme spatial outliers. Both physiological phenotypes were independent of active psychostimulant use. LimitationsWhile this study included a large and diverse group of children, the eye-tracking data were collected during a brief resting period -- watching a simple cross on a screen -- which may not capture how children behave in everyday, real-world situations. Because holding still for eye-tracking can be difficult, particularly for children with more severe symptoms, some data were lost; however, we statistically accounted for how much data each child contributed. Finally, while we confirmed that ADHD medication taken on the day of testing did not explain our findings, complete medication records were not available for every participant in this large observational study. ConclusionsPupillary dynamics and oculomotor stability associate with the ASD-ADHD spectrum through differing analytical patterns during resting states. Baseline autonomic volatility is more strongly captured by dimensional models of autistic trait severity, whereas baseline gaze instability is more clearly differentiated across categorical diagnostic groups, exhibiting a sub-additive interaction in comorbidity. Integrating both dimensional and categorical frameworks provides a more comprehensive understanding of these physiological variations, establishing a necessary foundation for future naturalistic and sensory-evoked research.