This paper shows the experimental setup and preliminary results of a work in progress that looks to induce and recognize the emotional influence of a character. The experiment starts with a user seated on an bus. A character takes the bus, and he walks and sits in front of the user. Three different behaviors are shown by the character, i.e., indifference, anxiety, and aggressiveness. A set of sensors were used to record the biometric signals of the user, and a bag of features approach is applied to the signals in combination with a machinelearning algorithms to recognize different emotional responses.