Directional information flow analysis in memory retrieval: a comparison between exaggerated and normal pictures.

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Machine Learning

Background:

• Working memory is crucial for learning but has limitations.
• Cognitive tasks can enhance working memory.
• Understanding information flow in visual working memory is key.

Purpose of the Study:

• Investigate information flow during a visual working memory task with exaggerated vs. normal pictures.
• Analyze electroencephalogram (EEG) signals using adaptive directed transfer function (ADTF).
• Examine network connectivity and P300 sub-components during retrieval.

Main Methods:

• Recorded EEG signals from 10 healthy men during encoding and retrieval tasks.
• Applied ADTF to event-related potentials (ERPs) for dynamic analysis.
• Used Wilcoxon test and five classifiers for feature selection and classification.

Main Results:

• The Z-values of Ge were most distinctive for network properties.
• K-nearest neighbors (KNN) classifier achieved 81% accuracy, 77% F1-score, and 81% specificity.
• Information flowed continuously in the right hemisphere from P3a to P3b for exaggerated pictures.

Conclusions:

• Exaggerated pictures elicit greater activation in attentional networks during visual working memory retrieval.
• These findings suggest exaggerated stimuli capture more attention, demanding increased cognitive resources.
• ADTF and machine learning effectively differentiate between exaggerated and normal picture retrieval.