Background: Educational psychology research has linked fluid intelligence (Gf) with working memory (WM), but it is still dubious whether electroencephalography (EEG) markers robustly indicate Gf. This study addresses this issue and notes the relationship between WM task-related EEG markers with Gf and academic performance.
Method: A sample of 62 healthy children between the ages of 9 and 12 years was selected to perform three tasks: (1) Raven's Standard Progressive Matrices (RSPM) test to assess Gf; (2) 2-back task to assess central executive system (CES); and (3) delayed match-to-sample task to assess short-term storage. These subjects were divided into high ability (HA) and low ability (LA) groups based on their RSPM scores. Support vector machine and logistic regression were used to train the EEG candidate indicators. A multiple regression was used to predict children's academic performance using P3 amplitude, P2 latency, and θ-ERS.
Results: Behavioral results demonstrated that the correct rate of the HA group is higher than that of the LA group. The event-related potential results of the 2-back task showed that the P3 amplitude of the HA group was relatively larger and that the P2 latency was shorter than that observed in the LA group. For the delayed matching to sample task, the θ-ERS of the LA group was higher than that of the HA group. However, the area under the curve of these three indicators for Gf was < 0.75 for each and < 0.85 for the combined indicators. In predicting academic performance, only P3 amplitude showed a significant effect.
Conclusion: These results challenge previous findings, which reported that P3, P2, or theta power might be used in standard psychometric tests to assess an individual's intelligence.
Keywords: academic achievement; event-related potentials; event-related synchronization; fluid intelligence; machine learning.
Copyright © 2020 Luo and Zhou.