Despite the interest in the neuroimaging of working memory, little is still known about the neurobiology of complex working memory in tasks that require simultaneous manipulation and storage of information. In addition to the central executive network, we assumed that the recently described salience network [involving the anterior insular cortex (AIC) and the anterior cingulate cortex (ACC)] might be of particular importance to working memory tasks that require complex, effortful processing. Method: Healthy participants (n = 26) and participants suffering from working memory problems related to the Kleine–Levin syndrome (KLS) (a specific form of periodic idiopathic hypersomnia; n = 18) participated in the study. Participants were further divided into a high- and low-capacity group, according to performance on a working memory task (listening span). In a functional magnetic resonance imaging (fMRI) study, participants were administered the reading span complex working memory task tapping cognitive effort. Principal findings: The fMRI-derived blood oxygen level dependent (BOLD) signal was modulated by (1) effort in both the central executive and the salience network and (2) capacity in the salience network in that high performers evidenced a weaker BOLD signal than low performers. In the salience network there was a dichotomy between the left and the right hemisphere; the right hemisphere elicited a steeper increase of the BOLD signal as a function of increasing effort. There was also a stronger functional connectivity within the central executive network because of increased task difficulty. Conclusion: The ability to allocate cognitive effort in complex working memory is contingent upon focused resources in the executive and in particular the salience network. Individual capacity during the complex working memory task is related to activity in the salience (but not the executive) network so that high-capacity participants evidence a lower signal and possibly hence a larger dynamic response.

Mathematics is an important skill that is taught to all children in the UK in a structured manner from a very early age. The purpose of this thesis was to examine how working memory (Baddeley & Hitch, 1974a; Baddeley & Hitch, 1994) and UK curricular mathematics are related, if specific components of working memory were more impactful upon performance in mathematics than others, and if we can predict mathematics outcomes using working memory measures. With reference to the influence of working memory on overall curricular mathematics performance, a cohort of 70 children from two primary schools in the North West of England was tested annually from their Reception year (mean age 5yrs 1m) at school to Year Two (mean age 6yrs 11m ). The study used a number of working memory tasks, a UK curricular mathematics test, and two Performance Measures. This allowed data to be analysed both in a cross-sectional manner and longitudinally (Chapter 5).The thesis also differentiates UK curricular mathematics into four separable “strands”, Number, Calculation, Measures, Shape and Space, and Problem Solving. These strands are described consistently throughout the UK mathematics curricular literature (DfEE, 1999; DfEE & QCA, 1999a; DfES, 2003a) and the cohort data was used to statistically analyse the relationships between working memory and each strand in turn using a correlational design in Chapters 6 to 9.Results indicated that working memory is a robust predictor of overall mathematics performance (Chapter 5), and of the Calculation Strand (Chapter 7). This finding was demonstrated in both the cross-sectional analyses and also in the longitudinal regression analyses. Of the working memory measures a distinct pattern of association was revealed. In particular the data imply that there is a strong role for the central executive at each age range, but in Year One verbal short-term memory emerges as an important predictor variable. Working memory also showed significant predictive influence over the remaining three curricular mathematics strands that were measured, particularly at the youngest age grouping, but working memory was not found to be a robust longitudinal predictor of Number, Problem Solving or Measures, Shape and Space. The overarching conclusion is that working memory, and in particular the central executive, may support the development of early curricular mathematical skills independent of the influence of age and Performance Measures. The practical and theoretical implications are considered.