Combining virtual reality and physiological measurements to assess emotions triggered by fragrances
89
Presented by: Xhino MELEQI
Introduction: Analysis of human physiological reaction to stimuli is a topic of interest for many disciplines and companies. Identification of emotion remains challenging due to: i) difficulty of labelling an odour-induced emotion, ii) absence of a references database that can be used to associate the obtained physiological data with emotional label. We have developed a new methodology that identifies the emotion induced by an odour, combining self-assessment survey with 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 opens the way to a precise labelling of the odour-induced emotions.
Method: 30 participants with no smell disfunction (mean age 37,6 +/- 10,4 , 26 women) have been recruited. They were stimulated by 10 selected raw materials, 30 perfumes, and 20 movies in virtual reality. Physiological data such as heart rate, skin conductance and respiratory volume were collected from participants. Self-assessment questionnaires were also presented. Participants had to evaluate valence, arousal, intensity and naturality of each stimulation. Principal component analysis (PCA) was run for both olfactory and visual stimulations. Finally, stimulations were clustered using k-means analysis.
Results: The principal component analysis of odour and videos reaches a score of 68,55% on the 2 mains axis. The same analysis on ingredients and videos reaches a score of 67,18% on the 2 mains axis. The videos associated with the same emotional term are grouped by principal component analysis as well as by k-means analysis. Perfumes, raw material and videos are mixed in different groups associated to 8 emotions.
Discussion: Results suggest that perfumes and raw materials generate emotions similar to virtual reality. Videos with same emotional term have been associated in the PCA as well as in k-means analysis. Those results suggest that in a cluster formed by videos, olfactory stimulations (perfumes or ingredients) trigger similar emotions. Thus, every raw material and perfumes have been associated with at least one emotional term of reference.
Conclusion: This methodology implicating virtual reality, allows to easily create emotional references database. Finally, our protocol opens the way to relatively fast odour-induced emotions evaluation based on a combination of subjective and objective criteria.
Method: 30 participants with no smell disfunction (mean age 37,6 +/- 10,4 , 26 women) have been recruited. They were stimulated by 10 selected raw materials, 30 perfumes, and 20 movies in virtual reality. Physiological data such as heart rate, skin conductance and respiratory volume were collected from participants. Self-assessment questionnaires were also presented. Participants had to evaluate valence, arousal, intensity and naturality of each stimulation. Principal component analysis (PCA) was run for both olfactory and visual stimulations. Finally, stimulations were clustered using k-means analysis.
Results: The principal component analysis of odour and videos reaches a score of 68,55% on the 2 mains axis. The same analysis on ingredients and videos reaches a score of 67,18% on the 2 mains axis. The videos associated with the same emotional term are grouped by principal component analysis as well as by k-means analysis. Perfumes, raw material and videos are mixed in different groups associated to 8 emotions.
Discussion: Results suggest that perfumes and raw materials generate emotions similar to virtual reality. Videos with same emotional term have been associated in the PCA as well as in k-means analysis. Those results suggest that in a cluster formed by videos, olfactory stimulations (perfumes or ingredients) trigger similar emotions. Thus, every raw material and perfumes have been associated with at least one emotional term of reference.
Conclusion: This methodology implicating virtual reality, allows to easily create emotional references database. Finally, our protocol opens the way to relatively fast odour-induced emotions evaluation based on a combination of subjective and objective criteria.