Introduction: The analysis of human physiological responses to stimuli is a topic of interest to both academic and industrial researchers. Emotion identification remains a challenge due to: i) the difficulty of labeling an odor-induced emotion, ii) the lack of a reference database that can be used to associate physiological data with the emotional label. We developed a new methodology based on the combination of a self-report survey and physiological analysis. The identification of an emotion is determined by a comparison of olfactory and visual responses, the latter being previously associated with an emotion. Our protocol paves the way for accurate labeling of odor-induced emotions.
Method: 30 participants without smell disorder (mean age 37.6 +/- 10.4, 26 females) were recruited. They were stimulated by 22 selected raw materials, 31 fragrances and 20 virtual reality movies. Physiological data such as heart rate, skin conductance, and respiratory volume were collected from the participants. Participants were also asked to rate the valence, arousal, intensity, and naturalness of each stimulation. K-means clustering analysis was performed with the olfactory and visual stimuli. Stimuli were clustered using k-means analysis.
Results: 13 relevant groups of responses were obtained. The intra-class variation is about 1.90 and the inter-class variation is about 6.43. Videos with the same emotional label were associated in the k-means clustering. Each cluster is defined by at least one emotional term. Olfactory stimulations are classified in those groups.
Discussion: The results show that the response patterns following an olfactory stimulation are comparable to those obtained during virtual reality stimulation. We were also able to highlight the importance of the subjective perception of the stimulation on the responses and finally on the triggered emotion.
Conclusion: This methodology involving virtual reality allows to easily create databases of emotional references.