I remember when I watched Mr. Sapolsky's lecture on the limbic system a month ago. It was so fascinating to know that the understanding of the limbic system was preceded by the understanding of olfactory bulb in rats. He then said that "....not an emotion on earth that's not intertwined with olfaction". Reading many papers and book in the area of affective computing made me overconfident that I might be the groundbreaker person that came out with the idea of automatic emotion detection that would take smell as a feature. This is because not a single research that I have read has taken into account the smells to predict emotions. Most of them talk about facial appearance changes that can imply emotions or about natural language, either they are written or spoken. However, lately, I realized that I might not be the first one that invented this idea. I, in the other hand, started to think that maybe it is not that none has the same idea; maybe it is just because the tools the research will depend on has not been fully developed yet.
While current technology can imitate our vision sense in term of getting pictures and even object detection has been more and more advance in the Computer Vision, it is not the case with technology for smells digitation or detection or simply sensor. Via Ted Talk given by Johannes Bintinger, the chemoreceptors that are needed to sense the smells are in the stage of early development. Bintinger said that we have 300 different olfactory receptors (or 40 million different olfactory receptor neurons as said by Ted ed or 400 receptors as said by Aroma Bit CEO) to differentiate 50 thousand different odors, yet they are still hard to develop. There is a Tokyo-based company that also developed a smell sensor called Aroma Bit. Nevertheless, only a very small number types of sensor that have been developed and the objects that are going to be smelled have to be pretty close to the sensors.
Then, maybe you question me: so, what is your idea when the dependency is fully available? Based on James theory of emotion, our emotions are simply produced by the external input that we sense. With this, we can assume that knowing the smells and odors in the environment can lead to an understanding of the emotions of the people nearby. The newer theory says that the emotion elicitations are not merely influenced by our external perceptions, it is also influenced by the internal state like our knowledge/experiences plus our appraisal system. Hence, different input data can result in different emotions in different people. Nonetheless, knowing the smell might give us knowledge that can be used to predict emotions, even though we should better consider other knowledge.
I have another hypothesis on how we can use the smell technology to predict emotions. First, if the theory of mind has something to do with olfaction --that we can use our smell sense to predict others' emotions in our daily life like when we also use it to recognize our relatives-- then maybe we can use the smell sensors to do the same thing in the future. BTW, if you have some papers or information about the relationship between the theory of mind with our olfactory system, would you please let me know?
For now, I maybe cannot do the research that use the olfaction to predict humans emotions. In the future, I guess I will stay stick to this problem and wait for the tools to be sufficiently developed. Meanwhile, I guess I am going to do another viable research in the same area.*Disclaimer: the cover pictures in this site are from sources that declare the picture as are free to use.