The way autistic persons process information in different areas of experience is far from being understood. The present paper discusses the idea of regularity detection and its relation to theories of autism and different autistic symptoms. The relevance of such a concept for autistic information processing is further explored in an artificial neural network simulation (using a simple recurrent network) which analyzes the impact of disrupting learning by increasing the network learning rate. Results are discussed in relation to weak central coherence theory and different aspects autistic behavior.
Keywords: autism, regularity detection, central coherence theory, artificial neural networks