Objectives: To develop a functional hypothesis and a neural-network model at the synaptic level to account for the evidence for hyper-perception and perceptual instability in autism.
Methods: Inspired by the analysis of individual cases, we developed a recurrent cortical network model that consists of excitatory and inhibitory populations, interconnected with synaptic connections with slow activity-dependent modifications in the intrinsic connections of the network leading to a changing balance between excitation and inhibition.
Results: Overall, the cases we studied suggest that severe autism is characterized by stimulus driven, winner-takes-all type of sensory processing in which a strong stimulus tends to extinguish other stimuli, leading to intermittent perceptual collapses. These collapses could be avoided by various detaching strategies such as looking to the side or covering the ears. Our preliminary investigation of the model shows that intermittent hyper states could be accounted for by abnormal synaptic homeostatic plasticity that normally regulates the balance between excitation and inhibition in cortical columns, and its failure could lead to unstable activity and intermittent runaway excitation at different levels of cortical processing, including perceptual sensitivity, attention and arousal.
Conclusions: Analysis of cases of severe autism and investigation of a simple neural-network model suggest that a local fault in homeostatic plasticity could lead to intermittent runaway cortical excitation, which may correspond to perceptual collapses. These could be reduced by detachment, thus initiating a severely abnormal developmental process.