A network analysis of executive functions before and after computerized cognitive training in children and adolescents

Abstract

Executive functions (EFs) play a key role in cognitive and socioemotional development. Factor analyses have revealed an age dependent structure of EFs spanning from a single common factor in early childhood to three factors in adults corresponding to inhibitory control (IC), switching and updating. IC performances change not only with age but also with cognitive training. Surprisingly, few studies have investigated training-related changes in EFs structure. We used the regularized partial correlation network model to analyze EFs structure in 137 typically developing children (9–10 years) and adolescents (15–17 years) before and after computerized cognitive training. Network models (NMs) —a graph theory-based approach allowing us to describe the structure of complex systems— can provide a priori free insight into EFs structures. We tested the hypothesis that training-related changes may mimic developmental-related changes. Quantitative and qualitative changes were detected in the EFs network structure with age and also with cognitive training. Of note, the EFs network structure in children after training was more similar to adolescents’ networks than before training. This study provided the first evidence of structural changes in EFs that are age and training-dependent and supports the hypothesis that training could accelerate the development of some structural aspects of EFs. Due to the sample size, these findings should be considered preliminary before replication in independent larger samples.

Latent network analysis of executive functions across development

“…Should this hypothesis gain support over the idea of a more integrated development of EFs, it would suggest that children are developing a more refined and complex set of cognitive skills, which has implications for interventions tailored to specific EF components. This differentiation through childhood, with increasing EF diversity with age, is also supported by EF studies using network models (Hartung et al, 2020;Karr et al, 2022;Menu et al, 2022;Younger et al, 2023). Network modeling, unlike SEM, allows for the study of relationships between variables without any a priori theoretical assumptions.…”

In search of better practice in executive functions assessment: Methodological issues and potential solutions.

“…Finally, network modeling, which proposes that cognitive processes are conceptualized as networks of directly related manifested variables, has been proposed to overcome the limitations of reflective and formative models (Schmittmann et al, 2013). Interestingly, network modeling has proven to be an effective tool to study the differentiation process of executive functions during development (Hartung et al, 2020; Karr et al, 2022; Menu et al, 2022).…”

Development of Executive Functions' Structure in 8-12 Year Olds

“…Finally, the models proposed are strongly influenced by the CFA approach proposed by Miyake et al (2000), which has dominated the field during the last decades. However, we recognize that other modeling approaches exist, which have already contributed to improve our understanding of the development of executive functions' structure, such as exploratory SEM or network modeling (Hartung et al, 2020;Karr et al, 2018;Menu et al, 2022).…”