Mining EEG with SVM for Understanding Cognitive Underpinnings of Math Problem Solving Strategies

Area of Science:

• Neuroscience
• Cognitive Science
• Computer Science

Background:

• Understanding the neural basis of mathematical problem-solving is crucial for cognitive science.
• Existing methods for analyzing brain activity often lack the precision to capture complex cognitive strategies.

Purpose of the Study:

• To develop and validate a new methodology for extracting patterns from brain electric activity using data mining and machine learning.
• To identify specific brain network characteristics associated with distinct mathematical problem-solving strategies.

Main Methods:

• Collected electroencephalographic (EEG) data during math problem-solving tasks.
• Utilized data mining and machine learning for pattern extraction, focusing on correlations and phase synchronization between EEG channels.
• Employed feature selection to identify relevant brain functional network features and connections.

Main Results:

• Developed a robust methodology for analyzing brain electric activity.
• Identified a suitable brain functional network size relevant to math problem-solving strategies.
• Discovered the most relevant connections within the brain network, distinguishing between effective and ineffective strategies.

Conclusions:

• The proposed methodology effectively extracts meaningful patterns from brain activity related to cognitive processes in mathematical problem-solving.
• This approach can help elucidate the neural underpinnings of mathematical cognition and inform educational strategies.