Proceedings of the Annual Meeting of the Cognitive Science Society

Why do neural responses decrease with practice? We used a predictive neural network model of sentence processing(St. John & McClelland, 1990) to simulate neural responses during language understanding, and examined the model’s correlateof neural responses (specifically, the N400 component), measured as stimulus-induced change in hidden layer activation, acrosstraining. N400 magnitude first increased and then gradually decreased over training while comprehension performance atthe output steadily rose with practice. These results fit the developmental trajectory of N400 amplitudes. Importantly, theyalso address the reduction of neural activation with practice. In the model, the reduction is due to continuous adaptation ofconnection weights over training. As connection weights between hidden and output layer grow stronger, less hidden layeractivation is necessary to efficiently modulate the output. This shift of labor from activation to connection weights might be animportant mechanism contributing to the reduction of neural activation with practice.