Revisiting working memory: Are domain, process and global models mutually exclusive, nested or orthogonal?
Abstract
Working memory (WM) is a cognitive function whereby task-relevant information is actively maintained and manipulated in mind for goal-directed behaviour. Three competing models, here dubbed the global, domain and process models, have attempted to explain its neural underpinnings. Despite extensive research however, no consensus has been reached. Here, we use two new WM paradigms to demonstrate that all three models are partially correct. In the first experiment, our results show that selected frontoparietal regions (MD), from the global model, are largely stimulus-independent. However, more posterior and caudal frontoparietal regions show stimulus-dependent activations as described by the domain model. In the second experiment, our results reveal that a dorsal MD sub-network is more active when information is manipulated, as described by the process model. Thus, WM is best represented by all three models, with the process model nested within the global, and the domain model partially independent from the others.