Why working memory measures 'work': Proactive interference in tests of immediate memory

Working memory (WM) span tasks (e.g., operation span; OSPAN) “work” because they reliably predict higher-order cognitive ability (e.g., Ravens Advanced Progressive Matrices; RAPM). Surprisingly, susceptibility to proactive interference (PI) may account for co-variation between WM and RAPM. Experiment 1 consisted of RAPM and multiple versions of OSPAN and a cued recall task (CR), modified for the category of the to-be-remembered stimuli (digits or words). Switching within-task from digits-to-words and back provided release from PI, while continued exposure to digits or words over multiple trials provided an opportunity for PI to build. Scores derived from PI-build trials, but not PI-release trials, correlated with RAPM and accounted for as much variance in RAPM as unmodified tasks. In Experiment 2, a distracter task (high or low difficulty) was added to CR and the processing component of OSPAN was manipulated for difficulty. OSPAN and CR “worked” when the processing component took controlled attention and compromised the ability to resist PI.