Paradoxical sensory reactivity induced by functional disconnection in a robot model of neurodevelopmental disorder

Hyper- and hyporeactivity to sensory stimuli is a diagnostic feature of autism spectrum disorder and has been reported in many neurodevelopmental disorders. However, the computational mechanisms underlying such paradoxical responses remain unclear. Here, using a robot controlled by a hierarchical recurrent neural network model with predictive processing and a learning mechanism, we simulated how functional disconnection alters the learning process and affects subsequent behavioral reactivity to environmental change. The results show that, through the learning process, functional disconnection between distinct network levels simultaneously lowered the precision of sensory information and higher-level prediction. These changes caused the robot to exhibit sensory-dominated and sensory-ignoring behaviors ascribed to sensory hyperreactivity and hyporeactivity, respectively. Furthermore, local functional disconnection at the sensory processing level similarly induced hyporeactivity due to low sensory precision. These findings suggest a computational explanation for co-existing sensory hyper- and hyporeactivity and insights at various levels of understanding in neurodevelopmental disorders.