摘要

Observations of higher dual-task costs for within-domain than cross-domain task combinations constitute classic evidence for multi-component models of working memory (e.g., Baddeley, 1986; Logie, 2011). However, we report an asymmetric pattern of interference between verbal and visual-spatial tasks, such that imposing a verbal memory load provokes graded decreases in visual memory performance, but imposing a visual memory load does not much affect verbal memory performance. Across multiple experiments, we verify that this pattern cannot adequately be explained as a mere byproduct of stimulus recoding or strategic preference. Current working memory models do not predict this persistent finding, thus a change in ongoing debate about relationships between attention and maintenance of verbal and visual mental representations is necessary.