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 form of a 2D matrix where vertical dimension reflected number channels horizontal one—to time samples. focused on distinguishing before their behavioural manifestation; therefore, we utilized segments preceding response. To deal with data, included convolutional procedure transforming into 1D feature vector. introduced three types convolution, including convolutions along x- y-axes convolution both axes. As result, F1-score for was above 88%, which confirmed model’s ability predict errors EEG. Finally, discussed limitations our approach its potential use brain-computer interfaces prevent human critical situations.