Gains of item-specific training in visual working memory and their neural correlates.

Experts sometimes show higher working memory performance than novices but contrary to this finding, evidence for a positive effect of item-specific training is rare. This study provides evidence for item-specific training gains. We presented Chinese characters and artificial patterns (spotted figures) in a change detection task before and after training (varying set size from 1 to 3). A part of the Chinese characters were trained; others and the spotted figures were not trained. Memory capacity was between one and two items. For set size two, memory performance for trained characters was higher than for untrained characters and they were processed faster. Within superior intraparietal sulcus and middle occipital cortex (part of the putative posterior working memory network), the neural activity asymptotically increased with set size. Untrained items reached the activation maximum already at set size two. For this set size, the activity was significantly reduced for trained items so that a further increase from two to three items was observed. We interpret this difference as a correlate of a gain in neural efficiency. The size of this difference correlated with the training gain in memory. We assume that training causes a more efficient neural representation of trained items supported by long-term memory and this allows holding more items in working memory.