Odor perception is a complex phenomenon, but as objective or subjective as anything in perception, be it auditory, visual or tactile. Human olfaction is usually assumed as unreliable, highly subjective and - in our interpretation - looked down upon due to its bodily associations, biological decay and emotion.

We will show, in contrast, that through the combination of classic psychophysics experimentation and recent advances in AI it is possible to obtain results that include the prediction of how the odor of small molecules and mixtures is judged on arbitrary perceptual qualities such as 'musty', 'rancid', 'stale', 'putrid', etc., and that open-ended descriptions of olfactory perceptual experience can be used to diagnose diseases, covid infection in particular.

We will discuss the implications of these results for an understanding of olfactory perception as an interplay between its contextual/subjective and non-contextual/objective determiners, specifically language, perceptual learning and physicochemical molecular properties, as well as the parallels with similar conceptualizations of visual, auditory and tactile perception.