ABSTRACT
Mathematical and connectionist models represent the parallel architectures of neural networks. These models enable us to gain insight into how complex interactions within our cognitive processing systems may function. However, these simulation results may not necessarily match the neurobiological basis of cognitive mechanisms. Event Related Potentials (ERPs) on the other hand allow us to observe brain activity. In the case of language processing, the N400 component has proved to be associated with semantic processing. This chapter reviews evidence from ERP studies in both children and adults in relation to Adaptive Resonance Theory (ART), an artificial neural network model. ART neural network programs can help to provide an idea of how the developmental constraints at the neural level (e.g., strength of synaptic connections) may cause systemic constraints in word learning and categorization behaviors in infants and toddlers.
