Predicting Human Sense of Smell using Neural Activity in Mice:A Bio-mimetic Artificial Sensory Test Method

2015 Fiscal Year Final Research Report

• Project/Area Number: 25540125
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Kansei informatics
• Research Institution: Hiroshima University
• Principal Investigator: Tsuji Toshio 広島大学, 工学(系)研究科(研究院), 教授 (90179995)
• Co-Investigator(Kenkyū-buntansha): Takiguchi Noboru 金沢大学, 自然システム学系, 准教授 (20304462) ; Kurita Yuichi 広島大学, 大学院工学研究院, 准教授 (80403591) ; Soh Zu 広島大学, 大学院工学研究院, 助教 (80724351)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 嗅覚 / 官能検査 / ラット / マウス / 神経活動 / ニューラルネットモデル

Outline of Final Research Achievements

Currently, odors are generally evaluated using a sensory test method performed by
well-trained human panels. In this study, we propose an olfactory model to predict perceptual similarity between odors aiming for future development of an artificial sensory test device. The model was constructed based on the anatomical structure of the olfactory system in rats, and the neural activity patterns of rats was used to represent odorant inputs. The simulation results confirmed that the model has capability to predict the perceptual characteristics of mice. We also defined features that represent difference between the neural activity patterns, and compared to the perceptual similarity of human. The comparison results showed a certain correlation, and confirmed the neural activity of the rats can be used to predict perceptual similarity of human.

Free Research Field

ニューラルネット