An Adaptive Neural Network

For an advanced undergraduate/postgraduate course in departments of computer science, artificial intelligence and biology, this text looks at a new approach for modelling the adaptive capacities of neuronal circuits in the cerebral cortex. Types of behaviour considered are visual guidance of hand movement, invariant pattern recognition and language learning. The model shares features with connectionist models, but takes into account the hierarchical organizational levels of the brain: synapses, neurons, columns, maps and sensorimotor systems. The book proposes rules for local processing units formed by cortical columns. It covers network model linking columns, the ontogeny of learning abilities and mental representations. Overall it aims to develop a multi-disciplinary approach linking neurobiology, psychology, artificial intelligence and parallel distributed processing.