Quantified variation in repeated experiments according to representational similarity analysis

In cognitive science such as emotion processing, the subjects are required to repeat the experiment many times. However, less attention has been paid to the quantification evaluation of consistent variation. In this study, a pipeline is proposed to quantify and detect the significant difference that existed in repeated experiments. At first, a public electroencephalography (EEG) dataset is enrolled where the emotion tasks repeat three times for per subject. Then
$\alpha\beta\delta \gamma \delta$
band EEG signals are extracted from Butterworth bandpass filter. Later, differential entropy and representational similarity analysis are combined to detect variation from inter-subject similarity analysis. Inter-subject similarity matrix is adopted to quantify the difference in repeated experiments. Our preliminary results demonstrate that significant differences are detected in certain electrodes during the repeated experiments.