Updating and working memory training: Immediate improvement, long-term maintenance, and generalisability to non-trained tasks

Despite the popularity of working memory (WM) and updating training, recent reviews have questioned their efficacy. We evaluated a computer-based training programme based on the Running Span and Keep Track paradigms. We assigned 111 7-year-olds with poor WM and mathematical performances to updating training, one of the two control groups, or a fourth group, who were administered Cogmed, a commercially available programme. At the immediate posttest, updating training produced only marginal improvements relative to control, but this was sustained and became significant six months post-training. Cogmed training resulted in substantial improvement at immediate posttest, but became marginal at delayed posttest. Neither type of training resulted in better performance in mathematics or generalised to other WM tasks that differed more markedly from those used during training. These findings suggest that relations between WM or updating capacity and mathematics performance may be moderated by factors that do not benefit directly from improved capacity.