Categorical Dimensions of Human Odor Descriptor Space Revealed by Non-Negative Matrix Factorization

Abstract

In contrast to most other sensory modalities, the basic perceptual dimensions of olfaction remain unclear. Here, we use non-negative matrix factorization (NMF) – a dimensionality reduction technique – to uncover structure in a panel of odor profiles, with each odor defined as a point in multi-dimensional descriptor space. The properties of NMF are favorable for the analysis of such lexical and perceptual data, and lead to a high-dimensional account of odor space. We further provide evidence that odor dimensions apply categorically. That is, odor space is not occupied homogenously, but rather in a discrete and intrinsically clustered manner. We discuss the potential implications of these results for the neural coding of odors, as well as for developing classifiers on larger datasets that may be useful for predicting perceptual qualities from chemical structures.

Underlying dimensions in the descriptive space of perfumery odors: Part II

“…Given such disparity and the fact that odor percepts depend on prior learning and experience (Wilson and Stevenson, 2003;Li et al, 2006), many olfactory researchers even doubt if it is possible to reach a standard map of scents. Actually, as the olfactory descriptive space is structured in at least 10 dimensions (Castro et al, 2013), it is uncertain how to project such space over a two-factorial plane. Nonetheless, fragrances are basically cosmetic pleasant scents and, hence, many food flavors are rarely present (e.g.…”

Iran Smell Identification Test (Iran-SIT): a Modified Version of the University of Pennsylvania Smell Identification Test (UPSIT) for Iranian Population

“…In order to replace unfamiliar odors, they were also asked to propose some odors which people from different regions of Iran are commonly encountered in their daily life. By considering the odor categories proposed by Castro et al (2013) and feasibility of supplying the odorants, the first version of Iran-SIT was designed using 40 items.…”

“…There are thousands of aromas that humans can smell. Castro et al (2013) have used a computerized technique to whittle down odors to their most basic essence. They classified the odors into 10 basic categories; floral, fruity (non-citrus), woody, chemical, minty, sweet, pungent, popcorn, citrus, and decayed.…”

The number of olfactory stimuli that humans can discriminate is still unknown

“…These maps are characterized by the attribute that pairs of items which are considered intuitively to be perceptually near (rated similar or difficult to discriminate) are nearer to one another on the map than pairs of items which are perceptually more distant (rated dissimilar or easy to discriminate). There are many algorithms for generating such maps, many of which have been used before in olfaction, including variants of PCA (Zarzo and Stanton, 2006;Khan et al, 2007;Koulakov et al, 2011), non-negative matrix factorization (NMF, [Castro et al, 2013]), and multi-dimensional scaling (Mamlouk et al, 2003). While there are open questions in the generation of these maps (e.g., how many dimensions should they have?…”