AN ARTIFICIAL NEURAL-NETWORK ANALOG OF LEARNING IN AUTISM

An artificial neural network is simulated that shares formal qualitative similarities with the selective attention and generalization deficits seen in people with autism. The model is based on neuropathological studies which suggest that affected individuals have either too few or too many neuronal connections in various regions of the brain. In simulations where the model was taught to discriminate children with autism from children with mental retardation, having too few simulated neuronal connections led to relatively inferior discrimination of the two groups in a training set and, consequently, relatively inferior generalization of the discrimination to a novel test set. Tao many connections produced excellent discrimination but inferior generalization because of overemphasis on details unique to the training set. It is concluded that within the context of the current model, the neuropathological observations that have been described in the literature are sufficient to explain some of the unique pattern recognition and discrimination learning abilities seen in some people with autism as well as their problems with generalization and concept acquisition. The model generates testable hypotheses that have implications for understanding the pathogenesis, treatment, and phenomenology of autism.