Due to the role of emotions in our daily lives, an abundance of scientific research has emerged that seeks to answer questions about our emotional landscapes through our body’s physiological responses. The field of recognizing emotion from facial expressions with the help of computer vision has also developed in the last two decades, In the challenge of recognizing human emotions, it makes sense to use a multimodal approach that can emphasize the advantages and reduce the disadvantages of each method. However, little attention is paid to the influence between various methods. In this master's thesis, we research the influence of intrusive measuring on the emotional expression of the face. In the theoretical part of the thesis, we first focus on the various theories of emotions and continue with a review of a wide range of methods for measuring and eliciting emotions in people. Before delving into the empirical work, we look at the poorly researched phenomenon of measuring anxiety and the classification of intrusiveness of measuring methods. In the empirical part, we describe our study, whose goal was to assess the impact of intrusiveness on the facial expression of participants using the Noldus FaceReader program. 44 participants were filmed via hidden camera while watching a protocol with ten emotion elicitation videos from the Samson et al. (2016) collection. After each video, the participants completed a nine-level SAM questionnaire for self-assessment of valence and arousal. The recordings of the participants were then analyzed with the FaceReader emotional state recognition program. The control group had no other conditions, while in the experimental group, we added increasingly intrusive measuring methods to the participants. The results of the groups were compared with one another. We hypothesized that the facial expressiveness of the experimental group would decrease more during the experiment than the control group. The results of the analysis with FaceReader showed the opposite - the program measured higher levels of expression in the experimental group than in the control group, which we prescribe to a difference of preferences in groups. Increased intrusiveness did not cause any noticeable differences. In our conclusions, we explain the results of the analysis with FaceReader and list other findings during our research. Also, we offer further research ideas, improvements for the current experiment, and highlight good practices for all further researchers of the field.
|
