Cognitive Analysis of Mental States of People According to Ethical Decisions Using Deep Learning Approach

Human behavior is a complex action which has provoked the thoughts of many people for a long time. However, very little about the cognitive aspect of an individual person's personality is known till the date. In order to differentiate people with significant differences in personality from their brain responses, we must at first be able to classify them based on their thought processes. From the viewpoint of classical ethics, people can broadly be classified into two main classes, namely, Categorical and Consequentialist. In this paper, we conduct several experiments where the subjects experience various ethical dilemmas and their ethical values in response to the presented stimuli are investigated through a question-answer session. The brain responses of the subjects are acquired using electroencephalography, which is then fed to an attention based parallel Convolutional Bi-directional Long Short-Term Memory (AConvBi-LSTM-NN) network with an ultimate aim to classify people into two above mentioned categories.Apart from the application point of view, the novelty of the paper lies in representing the EEG time-series into a sequence of multispectral 2D images which contain the spatial information of the acquired signal. The spectral information, along with the EEG time series (temporal information) are then used to train the proposed convolutional Bi-LSTM Network. The experimental results demonstrate promising results in classifying people based on their ethical values from their brain responses, with high classification accuracy. This provides scope for a new direction of research which can be further explored.