It has been suggested that diverse interventions applied within children's everyday contexts have the potential to improve working memory (WM) and produce transfer to real-world skills but little is known about the effectiveness of these approaches. This review aims to examine systematically the effectiveness of non-computerised interventions with 4-11 year olds to identify: (i) their effects on WM; (ii) whether benefits extend to near- and far-transfer measures; (iii) if gains are sustained over time; (iv) the active ingredients; and (v) the optimum dosage. Searches were conducted across 12 electronic databases using consistent keywords. Papers were screened by title and abstract ( 6212) and judged against pre-defined eligibility criteria ( 63). Eighteen papers were included in the review. They used a range of non-computerised WM intervention approaches that included: (i) adapting the environment to reduce WM loads; (ii) direct WM training with and without strategy instruction; and (iii) training skills which may indirectly impact on WM (physical activity, phonological awareness, fantastical play and inhibition). Both direct training on WM tasks and practicing certain skills that may impact indirectly on WM (physical activity, fantastical play and inhibition) produced improvements on WM tasks, with some benefits for near-transfer activities. The common ingredient across effective interventions was the executive-loaded nature of the trained task i.e., training on a task that taps into attentional and processing resources under executive control and not just the storage of information. Few studies reported dosage effects, measured far-transfer effects ( 4), or tested the durability of gains over time ( 4). The lack of a clear theoretical framework in many of the included studies resulted in ambiguous predictions about training and transfer effects, and inadequate use of outcome measures. Methodological issues also constrain the strength of the evidence, including: small samples sizes; an absence of blinding of participant and outcome assessors; and lack of active control groups. Further well-designed and controlled studies with clear theoretical underpinnings are required to expand and enhance the evidence base. The heterogeneity of the interventions and of the study designs (randomised and non-randomised) in the included papers limited the synthesis of evidence across studies. However, this diversity enabled the identification of key ingredients, notably the training of executive-loaded WM tasks, which can help inform novel approaches to WM intervention in everyday contexts.
