Abstract
While working memory (WM) is regarded as a capacity-limited process, theories differ on whether that capacity is set by modality-specific stores (Baddeley, 1986) or a single, unimodal store (Cowan, 1995). In support of the unimodal theory, several studies have observed dual-task interference between visual and verbal WM tasks (Morey & Cowan, 2004; Saults & Cowan, 2007). However, these studies used a visuospatial WM display that may be encoded as a set of propositional relations (e.g., blue on the left) and thus may rely on a verbal code. Here we re-examined dual-task costs in WM with a different spatial WM task that minimizes this potential confound. Participants were required to memorize the spatial positions of 1–5 dots. By always presenting stimuli along a line, the shape formed by the configuration of the dots was uninformative. Spatial memory was assessed by having participants judge whether a single-probe stimulus was presented in a location that matched one of the memory stimuli. In Experiment 1, participants performed the spatial WM task during the retention interval of a verbal WM task that required maintenance of 2–10 consonants. In contrast to previous studies, we found that the capacities for each of the two tasks were comparable in the single- and dual-task conditions. Additional experiments revealed no dual-task costs when an articulatory suppression task was added (Experiment 2), or with other auditory WM tasks (distinct, non-vocal sounds in Experiment 3 and bird songs in Experiment 4). To remove any possible contributions from sensory memory in these WM tasks, Experiment 5 extended the retention interval from 2s to 9s and Experiment 6 masked the sensory representations. Neither study demonstrated dual-task costs. Together, these results strongly suggest that the WM stores for auditory and visuospatial stimuli are independent, and support the existence of modality-specific WM systems.
This work was supported by NIMH grant (R01-MH70776) to R.M.