The Impact of Physical Activities on the Physiological Response to Emotions

Despite the advantages of using physiological sensors to collect emotion data, emotion recognition systems using physiological signals such as Electrodermal Activity (EDA), Electrocardiogram (ECG) or Electromyography (EMG) are mainly tested in controlled environments or under laboratory conditions. The use of physiological data in real-world scenarios has not been widely investigated. One of the main issues of using physiological data from real-world scenarios is that the data may also be influenced by movement and in some cases, the physiological response to emotions can be even confused with the one due to physical activities, such as walking or running. In this paper, we investigate the impact of physical activities in the recognition of emotions and provide new insights on how emotion data from physiological sensors are affected by these activities. We use two scenarios (one with and one without the influence of physical movement) to investigate the effect of physical activities in the Blood Volume Pulse (BVP) and the Skin Temperature (TMP) signals. To overcome these issues we used a random forest algorithm to model both scenarios. Our results show that by combining emotion data from both scenarios, we can achieve a recognition accuracy of up to 96%.