Saliency Models Reveal Reduced Top-Down Attention in Attention-Deficit/Hyperactivity Disorder: A Naturalistic Eye-Tracking Study.

Objective: Attention-deficit/hyperactivity disorder (ADHD) is a highly prevalent neurodevelopmental disorder. It is associated with deficits in executive functions, especially in visual attention. Deviant visual attention in ADHD is suspected to arise from imbalances between top-down and bottom-up mechanisms. However, it is unclear which of these mechanisms propels the aberrant visual attention. Method: In 815 medication-naïve children and adolescents (age range 5-21 years), differences in visual attention in participants with ADHD and neurotypical controls were investigated using eye tracking in a naturalistic video viewing task. Two opposing saliency models were used. Finegrained, based on low-level image features, was chosen to estimate bottom-up visually relevant areas. ViNet, a higher-level saliency model based on deep neural networks and trained on the gaze of neurotypical controls, was selected to determine top-down visually relevant regions. Correspondence between gaze and both saliency maps was calculated using normalized scanpath saliency, thus measuring the extent of coherence to bottom-up and top-down relevant contents. Results: Participants with combined ADHD showed lower mean normalized scanpath saliency for the top-down saliency map, but not the bottom-up one, compared with neurotypical controls. This contrast indicates poorer top-down control as a major contributor to impaired visual attention in combined ADHD. There was no significant effect for the predominantly inattentive ADHD group. Conclusion: This study demonstrated the use of eye tracking for differentiating between top-down and bottom-up visual attention. It shows that in combined ADHD, a reduction of top-down visual attention is key to an impaired competition between bottom-up and top-down visual attention.