Differences in fMRI activation between children with and without spelling disability on 2-back/0-back working memory contrast

Children (aged 10 to 12) with spelling disability (related to dyslexia) or with good spelling ability performed 2 fMRI nonverbal working memory tasks of comparable difficulty across groups in and out of the scanner-judging whether a pictured sea creature appeared two trials earlier (2-back) or was a target whale (0-back).The 2-back versus 0-back contrast captures ability of working memory to track changes over time. On this contrast, the good spellers and disabled spellers showed significant BOLD activation in many and generally the same brain regions. On group map comparisons, the good spellers never activated more than the disabled spellers, but the disabled spellers activated more than the good spellers in selected brain regions. Of most interest, 2 clusters of BOLD activation (distributed across brain regions) were observed in good spellers but 5 clusters were observed in disabled spellers. Within these clusters the good and disabled spellers differed in three regions (bilateral medial superior frontal gyrus, orbital middle frontal gyrus, and anterior cingulated), which are associated with cognition, executive functions, and working memory and were correlated with a behavioral spelling measure. Thus working memory is best described as a distributed architecture rather than a single mechanism; and good and poor spellers engage working memory architecture differently. We propose that spelling is an executive function for translating cognition into language (sounds and morphemes) and then into visual symbols rather than a mere transcription skill for translating words in memory into written symbols in external memory.