ODORSPACE · Predicting odor perception from odorant structure and neural activity in the olfactory system

The rules linking odor perception to odorant structure are unknown. No scientist nor perfumer can predict an odor based on its molecular structure, or decipher a molecular structure based on its smell. It is this puzzle we aim to solve. In vision and audition coding was probed by linking critical physical stimulus dimensions (wavelength/frequency) to patterns of neural activity. But what are the critical physical dimensions in olfaction? Scientists have probed this by linking restricted physico-chemical aspects of the stimulus, e.g., carbon chain-length, to neural activity. However, the olfactory system did not evolve to decode carbon chain-length, but rather to encode the world around us as revealed in olfactory perception. With this in mind we developed a novel perception-based olfactory space with tangible olfactory axes, based on statistical dimension-reduction of perceptual estimates obtained from humans. In Aim 1 we will test the hypothesis that our generated space predicts olfactory perception in humans. In Aim 2 we will test the hypothesis that our generated space predicts odorant-induced neural activity in olfactory cortex (using fMRI) and epithelium (using novel methods for measurement from human neurons in vivo, methods then further explored as a potential diagnostic tool for Alzheimer's disease). In Aim 3 we will test the hypothesis that our generated space explains neural activity previously measured in the olfactory system across species. In Aim 4 we will use this framework to tune an artificial nose for medical diagnostics. In vision and audition scientists can probe the system within agreed dimensions (color/wavelength; pitch/frequency). Similarly, our proposal generates an olfactory space where one can systematically probe molecular receptor tuning-curves, cellular spatial and temporal coding schemes, as well as higher-order perception. In other words, we propose a common framework for olfaction research.