Abstract:

The intent of teaching is to increase knowledge in long-term memory. To achieve this goal, it must be considered that human cognition has a very limited capacity for working memory when processing new information. Therefore, how to efficiently present information to reduce the load of working memory and promote the information from working memory into long-term memory is what the cognitive load theory focuses on (J. Sweller, 2007). After the American Association of Mathematics Teachers emphasized the importance of problem-solving in mathematics education, cultivating the mathematical problem-solving ability of primary and secondary school students has always been the goal of mathematics education in various countries (NCTM, 1989 & 2000). Research results show that experts are significantly better than novices in terms of problem-solving speed and correct rate of problem-solving (Larkin, 1985). In addition to the fact that experts have richer background knowledge than novices, another important factor is that experts have more problem schemas than novices.

Therefore, the aim of this study was to investigate how subjects’ behavior and electroencephalogram (EEG) performed in different types of mathematical questions between SAT high-performance group (HG) and SAT low-performance group (LG). We hypothesized the HG should have a lower cognitive load and should have lower mean amplitudes when answering questions. In this study, we used the SAT math test to identify HG and LG, a total of 75 high school students (HG=36, LG=39). There are three mathematical questions: Multiplication, Area, and Functions tests. Each condition has 60 questions, for a total of 180 questions. For EEG data, we analyze the EEG of subjects' answering by the response-locked event. Due to the outlier, and the problem of the noise signal, finally, the number of participants included for analysis is as followed: 62 (Multiplication test), 66 (Area test), and 63 (Functions test).

The behavioral results show that the accuracy of the Multiplication, Area, and Functions test was higher in the HG compared to the LG. The HG responded much faster than the LG did. For EEG data, we analyzed the mean amplitudes of the P7, P8, PO7, PO8, FZ, and OZ electrodes. The mean amplitude of PO7 was lower in the HG compared to the LG in the Multiplication test. The mean amplitudes of FZ and OZ were lower in the HG compared to the LG in the Area test. The mean amplitudes of P7, P8, PO7, PO8, FZ, and OZ were lower in the HG compared to the LG in the Function tests.

In conclusion, SAT high-performance group had a less cognitive load in answering mathematics questions. In addition, the FZ and OZ amplitude under the response-locked event may be a potential index for observing the cognitive load. The implication for education is that EEG results can broaden understanding of the answering mathematics questions.

Keywords:

Cognitive load, electroencephalogram (EEG), mathematics, educational neuroscience.