Relationships between Executive Function and Activity Monitoring in Children with and without ASD: Results from the ABC-CT Interim Analysis
Objectives: Explore the effect of EF ability on looking percent (head, background, overall looking) during an activity monitoring paradigm in ASD and TD participants above and beyond IQ and symptom severity (ADOS calibrated severity score).
Methods: Children six to twelve years old with Autism (n = 159) and typically developing individuals (n = 64) participated. An Eyelink 1000 Plus 500 Hz was used to track participants’ gaze while viewing several videos and images depicting two actresses engaged in a shared activity. Stimuli also included distractor objects placed throughout the scene and different gaze conditions (toward activity or the other actress). Trials were included if they contained valid data for more than 50% of the stimuli and if calibration error was less than 2.5 degrees. Percent time spent looking at heads (Head%), background (Background%), and the scene overall (TotalLooking%) was computed. The CASI-5 Inattentive and Hyperactive-Impulsive subscales were used as a proxy of EF.
Results: Correlations between eye-tracking outcome variables and CASI-5 EF subscales, partialling for IQ and ADOS symptom severity scores, were performed for each diagnosis group. No relationship was observed in the TD group. In contrast, Head% in ASD participants was significantly correlated with both Inattentive (r(155) = -.18, p = .024) and Hyperactive-Impulsive subscales (r(155) = -.23, p = .004), suggesting that as CASI-5 scores increased (indicating worse EF skills), time spent looking at heads decreased. A follow-up ANCOVA analysis looking at the relationship between Heads% and factorial effects of diagnosis, IQ, and CASI-5 was performed. This yielded a main effect of CASI-5 scores, F(1, 222) = 4.29, p = .039, and significant two-way interactions between CASI-5 and IQ and CASI-5 and diagnosis (F(1, 222) = 4.14, p = .043, F(1, 222) = 6.09, p = .024, respectively). The three-way interaction between diagnosis, IQ, and CASI was also significant, F(1, 222) = 5.51, p = .02.
Conclusions: Together, these analyses indicate that a complex relationship exists between CASI-5 and Head%. Although the nature of this relationship is currently unclear, the interactions suggest that CASI-5 is capturing variance in these social looking patterns that are unaccounted for by diagnosis and IQ alone. Thus, EF may play a role in social learning which may in turn impact future social skills. Interventions targeting EF domains may benefit both EF and social skills.