Interactions in the development of skilled word learning in neural networks and toddlers

Development is about change over time. Computational models have provided insights into the developmental changes seen in different cognitive phenomena, including within the domain of word learning. The present paper uses a computational model in tandem with a behavioral study to make and test predictions about the interdependencies between the emergences of different word learning biases. The model is used to investigate how the shape bias influences novel noun generalization to other types of items, and to guide a behavioral study of this effect in children. The results provide a novel view of biased word learning over time, and suggest that emerging biases interact with each other and influence how networks and children attend to different kinds of information over time.