What counts in the brain? The neural correlates of arithmetic in adults and children with and without learning disorders

Arithmetic constitutes a large part of our daily lives, yet its neural basis in adults and children with and without learning disorders is not yet fully understood. Neuroimaging research in adults has been mainly performed against the background of one theoretical framework: the Triple Code Model. This model acknowledges the influence of three codes, each with its own neural component: the magnitude code, located in the posterior parietal areas, is involved in estimating and comparing numerical magnitudes. The verbal code, which is localized in the left perisylvian areas, is implicated in the semantic representation of numerical magnitudes. Finally, the visual code plays a role in the visual processing of Arabic digits. Although the former two codes have been extensively researched, research on the role and the anatomical location of the visual code has been scarce to date. Using univariate and multivariate analyses, an fMRI study in adults revealed no focal region specifically hosting the visual code, yet that digits were represented as distributed patterns.

The Triple Code Model was based on research in adults, and is therefore not simply transferable to children. Neuroimaging research on arithmetic in children is relatively scare to date and has often focused on investigating the neural correlates of various strategies children use to solve arithmetic problems. However, these previous studies have been confounded by the use of, for example, different operations. An fMRI study in typically developing children in which we manipulated presentation format rather than operation indicated that children used procedural strategies when they were asked to perform non-symbolic subtractions, yet retrieved the solution from long term memory for symbolic subtractions. This study showed that neural strategy effects are not solely reliant on operation effects, but that they are dependent on the individual characteristics of arithmetic problems.

Despite the large role of arithmetic in everyday life, around one in ten children suffers from deficits in arithmetic processing (i.e., dyscalculia). Even more, children with reading disabilities (i.e., dyslexia) often show impairments in arithmetic as well. It remains however unclear if these comparable difficulties with performing arithmetic at the behavioral level, originate from similar neurobiological effects in dyscalculia and dyslexia. In an fMRI study in children with dyscalculia, children with dyslexia, children with comorbid dyslexia/dyscalculia and age-matched typically developing children, we found, using univariate analyses, hypo activation for all children with learning disorders compared to typically developing children. Furthermore, no brain regions were significantly more activated during arithmetic in children with dyscalculia compared to children with dyslexia, and vice versa. To rule out the possibility that power issues drove this null-effect, we used multivariate analyses that indicated that, despite clear differences in their behavioral profiles, all children with learning disorders were remarkably similar in terms of their neural profiles.

Collectively, these studies and the univariate and multivariate analysis techniques that were used, helped us gain more insight into what counts in the brain during arithmetic processing.