Resource-sharing in multiple-component working memory

Working memory research often focuses on measuring the capacity of the system and how it relates to other cognitive abilities. However, research into the structure of working memory is less concerned with an overall capacity measure but rather with the intricacies of underlying components and their contribution to different tasks. A number of models of working memory structure have been proposed, each with different assumptions and predictions, but none of which adequately accounts for the full range of data in the working memory literature. We report 2 experiments that investigated the effects of load manipulations on dual-task verbal temporary memory and spatial processing. Crucially, we manipulated cognitive load around the measured memory span of each individual participant. We report a clear effect of increasing memory load on processing accuracy, but only when memory load is increased above each participant's measured memory span. However, increasing processing load did not affect memory performance. We argue that immediate verbal memory may rely both on a temporary phonological store and on activated traces in long-term memory, with the latter deployed to support memory performance for supraspan lists and when a high memory load is coupled with a processing task. We propose that future research should tailor the load manipulations to the capacities of individual participants and suggest that contrasts between models of working memory may be more apparent than real.