Predicting Perceptual Decision-Making Errors Using EEG and Machine Learning

We trained an artificial neural network (ANN) to distinguish between correct and erroneous responses in the perceptual decision-making task using 32 EEG channels. The ANN input took the form of a 2D matrix where the vertical dimension reflected the number of EEG channels and the horizontal one-to the number of time samples. We focused on distinguishing the responses before their behavioural manifestation; therefore, we utilized EEG segments preceding the behavioural response. To deal with the 2D input data, ANN included a convolutional procedure transforming a 2D matrix into the 1D feature vector. We introduced three types of convolution, including 1D convolutions along the x- and y-axes and a 2D convolution along both axes. As a result, the F1-score for erroneous responses was above 88%, which confirmed the model's ability to predict perceptual decision-making errors using EEG. Finally, we discussed the limitations of our approach and its potential use in the brain-computer interfaces to predict and prevent human errors in critical situations.