Deciphering Odor Perception through EEG Brain Activity and Gas Sensors

Hsin-Ping Peng, Hao-Lung Hsiao, Chien-Hui Su, Yang Chen Lin, Po-Chih Kuo

Abstract: Recent technological advances have led to innovations like electronic noses and gas sensors, proficient in detecting distinct odors. Despite this, the field of AI and robotics has only marginally explored olfaction, a sense crucial for evoking emotions and memories. Our study investigates the correlation between gas sensor signals and EEG activity during odor recognition. By comparing our findings with questionnaire results, we suggest that individual experiences might influence odor recognition in the human brain. We designed an odor-dispensing system and recorded EEG responses from 15 subjects to six odors, alongside concentration data of four gases for each odor. These EEG and gas sensor data were analyzed using two neural networks for odor classification. Combining EEG and gas sensor data, we attained a 44% accuracy in 6-class odor discrimination, indicating the potential of this integrated approach as a unique ‘odor fingerprint’ for odor identification.