Preliminary comparative spectral analysis of EEG for two participants during a free recall working memory task

Poster Session F - Tuesday, April 16, 2024, 8:00 – 10:00 am EDT, Sheraton Hall ABC

Martin Pham¹ (martindopham@gmail.com), Hrishikesh Pable³, Robin Chhabra², Maryam Mehri Dehnavi¹, Amedeo D'Anguili²; ¹University of Toronto, ²Carleton University, ³Indian Institute of Technology Dharwad

We present exploratory analysis of EEG collected during a working memory task where (N=2) participants were presented with geometric shapes and words for free recall later on. We investigate the brain wave activities of different electrode regions associated with the default mode network (DMN) and task positive network (TPN) by comparing the complexity of channels with high coherence. The method of analysis is a multistep process: (1) compare the intra-subject wavelet coherence of all channels for each participant, (2) compute the multiscale entropy and fractal dimensions for groups of channels with high coherence, (3) use those complexity measures to predict the recall performance of participants as well as VVIQ scores that were measured separately. Measuring coherence in the wavelet domain allows for time-frequency analysis, separating the different brain wave bandwidths of interest during different (shapes and words) stimuli. A multiscale entropy method is used in order to measure the regularity of cohering signals across different resolutions and therefore bandwidths. Both Katz’s and Higuchi’s fractal dimensions are computed in order to measure the predictability and estimate the temporal dimensionality of signals. Although the indices do not show strong predictive value for recall and VVIQ scores, there is evidence of region-specific stimuli-specific brain wave activity during the memory encoding phase of the task.

Topic Area: METHODS: Electrophysiology