Retroactive interference with working memory consolidation: Visual, verbal, or central processing?

The speed by which information from vision is transformed into a working memory (WM) representation that is resistant to interference from ongoing perception and cognition is the subject of conflicting results. Using a dual-task retroactive interference (RI) approach, Nieuwenstein and Wyble (JEP:General, 2014) showed that WM consolidation is a slow process that continues for up to 1 s, even after a visual mask—contrary to previous, more rapid, estimates. Recent research using the RI approach has shown that the slow rate of WM consolidation is a structural limit (Carlos, Santacroce, & Tamber-Rosenau, OPAM, 2019). However, a major caveat to this research is that it has largely used visually-presented letter WM arrays. Thus, it is ambiguous whether slow consolidation is a visual WM, verbal WM, or central processing phenomenon. A single experiment by Nieuwenstein and Wyble (2014) demonstrated slow consolidation for a visuospatial WM item (an unfamiliar Kanji character), but the bulk of recent studies of WM storage capacity use color patches or other single-feature items. To determine if results from the RI/WM consolidation paradigm stem from the very same WM system as familiar storage capacity limits, we used a dual-task RI approach to evaluate the speed of WM consolidation for color patches. Replicating prior results with letter arrays, we observed an interaction between delay duration and the presence of a second task (p = .01, ηp^2 = 0.26). Moreover, when compared to letter array results, the delay x second task interaction was not different across WM array types (p = .44, ηp^2 = 0.03, BF10 = 0.150). The most parsimonious explanation of these results is that RI with WM consolidation stems from a central processing limitation that is involved in the consolidation of verbal, complex visuospatial, and simple color patch WM arrays.