critical Design Report So my initial idea was to use this sensor to detect people’s feelings while making choices. For example, using music or videos to guide users to meditate on a problem, (This problem should have two opposing preselected answers, such as: Should I choose to switch to a new job?) Then using sensors to monitor the real time fluctuations of the physiological signal. Then I try to represent these emotions with a simple one-dimensional value, which represents the intensity of the emotion. If the value is less than 0, it means it mainly remains negative emotion. If the value is greater than 0, it means there is positive emotion mainly. This is an illustration of the classification of emotions referring to the book "cognitive structure of emotion". Therefore, different objective conditions, behavior, and the brain's thinking process will trigger the release of different emotional signals (positive / negative) in the hypothalamus. So, when we trained the coded computer to detect whether these subconscious emotions are negative or positive when the user is concentrating on a choice problem in order to reversely infer their real choice. However, through experiments, I have found that this kind of sensor can't tell clearly the user's emotions (happy, angry, upset, joyful...), but only can detect the fluctuation of user's psychological state, just like the skin conductance sensor on lie detector can sense the speaker's psychological change, detect whether or not there is a sudden flurried of emotional change in order to determined whether he is lying. For example, ‘nervous’, ‘frightened’, ‘excited’ and ‘happy’ will all cause a sudden increase of the test curve, but it cannot be inferred whether this change is caused by positive or negative emotions. How to classify these data and classify emotions scientifically is very difficult. I did not find a reasonable solution, so I began to search for new methods of detection and expression.