Semantic knowledge can impair as well as benefit working memory performance: evidence of proactive facilitation and interference in children

Research with adults has led to the proposal of a gating mechanism between long-term memory and working memory, with information from long-term memory entering working memory only when it is beneficial. Across three experiments, we examine functioning of the gating mechanism in children using a novel cross-modal working memory task. In Experiment 1, 7-11 year-old children and adults were presented with four images of animals paired with auditorily presented sounds. Image-sound combinations were congruent with existing knowledge (e.g. cow-moo), incongruent (e.g. cow-baa) or involved neutral pseudowords (e.g. cow-jad). After a short delay, participants were presented with a sound and asked to report the paired image. Replicating previous research, adults showed superior accuracy for congruent pairings, with evidence of no difference between neutral and incongruent pairings. Children’s performance was qualitatively distinct, exhibiting both a congruency boost (congruent > neutral) and an incongruency cost (neutral > incongruent). Effects were replicated in children when set size was varied (Experiment 2). Experiment 3 then increased the interstimulus interval to account for children’s slower processing speed. Here, a congruency boost was consistently observed, with some evidence that an incongruency cost emerged at the lower set size. Taken together, these findings provide clear evidence that long-term memory benefits working memory performance in children. Crucially, however, children appear more sensitive to incongruent information from long-term memory relative to adults. We argue, therefore, that any gating mechanism is not fully developed in 7-11 year old children.