Long-Term Abacus Training Reshapes Substrates of Executive Function and Mathematics among Children

1109 

Abstract Submission 

Tianyong Xu¹, Xinyang Liu², Andrea Hildebrandt³, Changsong Zhou^1,4, Feiyan CHEN¹

¹School of Physics, Zhejiang University, Hangzhou, China, ²School of Psychology and Cognitive Science, East China Normal University, Shanghai, China, ³Department of Psychology, Carl von Ossietzky University of Oldenburg, Oldenburg, Germany, ⁴Department of Physics, Hong Kong Baptist University, Kowloon, Hong Kong

Tianyong Xu  
School of Physics, Zhejiang University
Hangzhou, China

Xinyang Liu  
School of Psychology and Cognitive Science, East China Normal University
Shanghai, China

Andrea Hildebrandt, Prof. Dr.  
Department of Psychology, Carl von Ossietzky University of Oldenburg
Oldenburg, Germany

Changsong Zhou, Prof. Dr,  
School of Physics, Zhejiang University|Department of Physics, Hong Kong Baptist University
Hangzhou, China|Kowloon, Hong Kong

Feiyan CHEN  
School of Physics, Zhejiang University
Hangzhou, China

Domain-general executive functions (EFs) serve critically for mathematical skills (Cragg et al., 2014), which relationship of these two closely-related abilities show a variable role among cohorts with individual difference (Best et al., 2011; Cragg et al., 2017; Spiegel et al., 2021; Dong et al., 2022). However, little is known about whether the plasticity of this relationship between EF and math as well as their individual difference during childhood can be affected by cognitive training and the neural underpinnings. By analyzing behavioral performances and brain functional neuroimaging of children with and without five-year abacus mental calculation (AMC) study, we investigated the training effect on the association and individual difference of EF and mathematics, together with relevant changes in neural substrates.

A total of 142 children were recruited in this experiment, whom were randomly assigned to the training group (n = 72) and the control group (n = 70). Children in the training group received a two-hour AMC training per week for five school years, while children in the control group did not have any prior experience in AMC. In the post-training session, all of them participated in the EF- and mathematics-associated behavioral tests. Among them, 66 children completed resting-state fMRI scanning performed on a 1.5T Philips MRI scanner. We made following investigations. First, we applied multiple group structural equation modeling (SEM) analysis to test group difference of EF-math relationship. Then, we calculated similarity of EF-math FC patterns for the overlap size of shared neural circuits. Further, we computed the inter-subject similarity of FC (ISFC) patterns in frontal-parietal regions. Finally, we explored the role of ISFC on the relationship between EF and mathematics by performing mediation analysis.

By demonstrating that training promoted EF and mathematical abilities (Fig. 1A), we also observed long-term abacus training reduced individual variability and enhanced the relationship between EF and mathematics (Fig. 1B, C). Using different constraint strategies, step-wise comparisons were made based on the SEM model, and all results in model comparisons were shown in the table (Fig. 1B). It revealed higher EF-math relationship (r = 0.856, p = 0.008) in the AMC group relative to the controls (r = 0.468, p = 0.060, Fig. 1C). These results demonstrated significantly decreased individual variances of mathematics and EF, as well as their enhanced relationship. Focusing on the relevant frontal-parietal regions from meta-analytic maps with EF/math-related terms in Neurosynth, we found that long-term training modulated FCs within these regions (Fig. 2A). Moreover, the AMC group presented higher overlapping circuits in EF-math FC patterns, which might support the enhanced EF-math transfer (Fig. 2B). Notably, training enhanced ISFC in frontal-parietal regions significantly (Fig. 2C), which partially mediated EF-related improvement in mathematical ability in the AMC group but not in the control group (Fig. 2D).

Our current study revealed the plasticity of transfer effect and individual difference affected by cognitive training. Long-term abacus training with visuospatial strategy can modulate FCs within frontal-parietal systems. Therefore, EF and mathematical abilities depending on the common neural substrates of frontal-parietal circuits could be improved and their relationship becomes strengthened. Besides, sustained training over multiple years with same strategy should reduce individual variability especially in the relevant neural networks. These different dimensions of training/learning as demonstrated here for EF and math is expected to be generalizable to other transferable abilities.

Executive Function, Cognitive Control and Decision Making

Higher Cognitive Functions Other ²

Skill Learning ¹

Connectivity (eg. functional, effective, structural)

BOLD fMRI

Other - Transfer effect, Brain plasticity, Cognitive training, Functional connectivity, Individual difference