Single Trial Classification of per-stimulus EEG between Different Speed Accuracy Tradeoffs Instruction

Area of Science:

• Cognitive Neuroscience
• Machine Learning
• Neuroimaging

Background:

• The speed-accuracy tradeoff (SAT) is crucial in cognitive processing, influencing decision-making speed and accuracy.
• Understanding distinct SAT strategies in neurological patients is vital for diagnosis and treatment.
• Neural mechanisms of SAT are studied, but classifying neural data for different SAT strategies is underexplored.

Purpose of the Study:

• To develop and validate a deep learning framework for classifying single-trial electroencephalography (EEG) signals based on instructed speed or accuracy prioritization strategies.
• To bridge the gap in classifying neural data for distinct SAT strategies.

Main Methods:

• A deep learning framework was implemented using a mirror-image judgment task dataset from 20 participants.
• EEG data were preprocessed and transformed using continuous wavelet transformation for time-frequency features.
• A channel-stacking technique converted EEG data into RGB-like images for input into a RegNet convolutional neural network for classification.

Main Results:

• The deep learning model achieved high classification accuracy in distinguishing between speed and accuracy strategies using single-trial EEG.
• The occipital region showed the highest classification accuracy (85.37%), followed by parietal (82.97%), frontal (80.46%), and central regions (78.57%).
• Ten-fold cross-validation confirmed the model's robust performance.

Conclusions:

• Single-trial EEG classification is feasible for differentiating between speed and accuracy cognitive strategies.
• This approach has potential applications in adaptive brain-computer interfaces and cognitive neuroscience research.
• Provides clinicians with a novel tool for real-time identification of cognitive strategies to tailor neurofeedback and rehabilitation protocols.