Induction method influence on emotion recognition based on psychophysiological parameters

Abstract

Recognizing emotions is an essential ability in our daily social interactions. However, there are individuals who have difficulties interpreting emotions, such as patients with autism spectrum disorders (ASD). In order to cope better with everyday life, emotion training can be a supporting factor for them. However, studies show that emotion training is not only helpful for patients with ASD, but also in the working environment, for example in trainings for managers or teams. In recent research, there are already approaches to use new technologies such as virtual reality to train emotional and social skills. For the evaluation of these new concepts, it is important to make the emotional state of a person measurable. Therefore, a measurement environment has already been developed at Furtwangen University. This is based on a multidimensional approach combining subjective and objective psychophysiological measures. Moreover, the development of facial emotion recognition (FER) systems based on machine learning techniques are also increasing for measuring a person's emotional state. Often, they focus on the recognition of Ekman’s basic emotions. To train and evaluate such FER systems, these basic emotions have to be induced in an individual. Therefore, a number of methods for emotion induction can be found in research, e.g. visual stimuli or mental methods. However, in most studies, only a few selected emotions, such as anger and happiness, were induced. Thus, there is a lack of studies that examined the induction of all six basic emotions.For that reason, the aim of the presented experimental study was to investigate two different methods of emotion induction for the six basic emotions anger, disgust, fear, happiness, sadness, surprise, and a neutral category. Overall, 14 women and 10 men (N = 24) aged between 19 and 59 years (M = 29.25, SD = 11.46) participated in the study. For the first induction method, affective visual stimuli from common emotional picture databases (EmoPicS, OASIS and IAPS) were used. For the second induction method, emotions were induced by a so-called autobiographical recall. Therefore, the participants had to imagine autobiographical situations that evoked the required emotion in them in the past. After each different induction of one of the six emotions or the neutral category, the participants’ emotional state was assessed using the two dimensions valence and arousal of the Self-Assessment Manikin (SAM). Furthermore, cardiovascular (ECG) and electrodermal (EDA) activity were recorded. The results show a significant interaction induction method x emotional category for both subjective assessments valence and arousal. Furthermore, based on the results of the psychophysiological responses of the participants (ECG and EDA), it is shown that the second method to induce emotions (autobiographical recall) was significantly more arousing than the first induction method using visual stimuli. To sum it up, the results of the experimental study show an influence of the induction method that is evident in both the subjective and the psychophysiological parameters.