Task base http www apexbt com
Task-based connectivity studies of verbal WM emphasized maturational changes of within and between network functional interactions. For example, van den Bosch et al. (2014) reported age-related functional connectivity changes in children and adolescents within frontoparietal, motor and cingulate networks. Using a similar task, Finn et al. (2010) found longitudinal decreases in connectivity between hippocampus and lateral PFC with age. Combined, these results emphasize the need to consider a broad range of areas and their functional interaction, rather than any single system. A central challenge for task-based functional magnetic resonance imaging (fMRI) is designing tasks that equivalently probe WM in different age groups (Church et al., 2010; Luna et al., 2010). Previous studies addressed this challenge by either statistically controlling for performance (van den Bosch et al., 2014) or matching performance, i.e., selecting low performers from adults to match the performance of children (Crone et al., 2006). However, statistically controlling for performance may hinder the ability to detect age effects (van den Bosch et al., 2014), as performance typically depends on age (see Satterthwaite et al., 2013b for an example of examining the unique effects of age and performance). Also, adults with low performance may not be appropriately representative. Not limited by tasks, resting-state fMRI (R-fMRI) is a powerful tool for mapping maturational changes in DMH-1 functional organization (Dosenbach et al., 2010) and indexing inter-individual differences in cognition and behavior (Kelly et al., 2008; Koyama et al., 2011). Most R-fMRI studies of verbal WM have focused on adults (Gordon et al., 2014; Jolles et al., 2013; Takeuchi and Kawashima, 2012). One study in children examined the maturation of functional connectivity underlying the improvement of cognitive control, a core component of manipulation (Barber et al., 2013). Those authors found that the intrinsic anticorrelation between the task positive and default network was greater in adults than children. In addition, the strength of this anticorrelation was associated with inhibitory control performance across groups. These results suggested that the development of this anticorrelation supports mature inhibitory control. However, this pioneering effort had a number of limitations: reliance on a seed-based correlation approach, treating age as a categorical variable, and lack of an adolescent group. Using R-fMRI, we systematically examined neural indices of WM performance in a cross-sectional developing sample (ages: 7–17yrs) utilizing a broad range of data-driven approaches. Compared to traditional seed-based correlation analysis, data-driven approaches do not require a priori hypotheses and allow for identifying previously overlooked brain–behavior relationships. Using a relatively large sample (n=68), we treated age as a continuous variable. WM was assessed using the Wechsler Intelligence Scale for Children (WISC-IV; Wechsler, 2004) Digit Span (DS) subtest, which includes: DS Forward (DSF) and DS Backward (DSB). These tasks were selected because they are well-validated measures commonly used in educational and clinical evaluation (Gathercole and Alloway, 2006), which increases the ecological validity of our findings. The available age-normalized scores also allow comparisons of performance across different ages. The cognitive literature has suggested that DSF and DSB rely on shared and distinct WM components (Hale et al., 2002; Reynolds, 1997). DSF is thought to depend on the ability to maintain information in the “phonological loop” and is strongly associated with language development (Baddeley, 2012). DSB has additional executive control requirements to transform and manipulate information (e.g., reverse the digit sequence). Thus, DSB is more reflective of cognitive control (Baddeley, 2012) and involves visual-spatial skills (St Clair-Thompson and Allen, 2013). Because the DS total (DST=DSF+DSB) score is widely used to index verbal WM abilities (Walshaw et al., 2010), we examined the aggregate, as well as distinct intrinsic brain correlates of DSF and DSB via two regression analyses: for aggregate analyses, we included DST as the variable of interest in a model; for distinct analyses, we included DSF and DSB scores in the same model to control the effect of the other.