Evaluating models of working memory: FMRI and behavioral evidence on the effects of concurrent irrelevant information