White is a mixture, made by a combination of signals at equal intensity across a perceptual space. White light can be split up into all the colors of the visible spectrum, and white noise covers a range of frequencies within the audible range.

Our other senses don't have as clearly defined ranges of perception. We can't give a smell, a taste, or a texture a number the same way that a color or a tone can be defined by a wavelength, but a fascinating recent paper shows that by mixing many different smelly molecules at equal intensities, our perception of the odor will converge on "olfactory white."

The researchers created this strangely neutral smell from different mixtures of up to thirty odors, chosen from a set of 86 molecules that represent a wide range of the kinds of things that we can smell. Human "olfactory stimulus space" contains thousands of molecules, from the fragrant and floral to the putrid. We can distinguish and name many smells, but odors don't map neatly onto a one dimensional spectrum. Sampling the multidimensional stimulus space of odors requires a much more complicated mapping of the smell universe. The figure on the left shows the position of the 86 molecules within two maps of olfactory stimulus space. The first is based on the way that we perceive odors (perceptual space, A) and the second based on the chemical structures of the molecules (physicochemical space, B).

The perceptual map is built with data from Dravnieks' Atlas of Odor Character Profiles of 144 different molecules. Each smell was compared by 150 professional noses against a list of 146 different odor descriptions like "fruity" "etherish" "decayed" or "seasoning for meat." This enormous multidimensional dataset represents the way that we experience and name olfactory space. Using principal component analysis this 146 dimensional space can be simplified into the two dimensional projection that represents the "distance" between the smells of the different odors in this space.

Physicochemical olfactory space is both more objective and more dense than the map of perceptual space, built from 1514 descriptions of the chemical structures of 1565 odorant molecules. These descriptions include things like the number of atoms, the presence or absence of specific chemical groups, and the weight and shapes of the molecules. The shape of a molecule isn't usually enough to predict its odor but there are some correlations that can be made between the 1514-dimensional space and other perceptual maps, based on professional descriptions or even brain activity.

The position of isoamyl acetate (banana oil) vs. skatole (a strong fecal odor at high concentrations) in different representations of olfactory space

Maps of brain, rodent olfactory bulb, or odor receptor activation in response to different odors represent a different kind of olfactory space, one mapped and coded in the structure of the sensory organs themselves. The activity map of the rat olfactory bulb for increasing concentrations of isoamyl acetate (banana smell) shows the regions of cells activating and sending signals to the brain saying "banana!"

I'm curious what these activity maps would look like for "white smell," although the authors suggest that the perception may not be caused by the wide activation that characterizes the perception of white light or white noise. Whereas white light equally activates the three color receptors of the eye, using electricity to stimulate the olfactory system leads to no sensation of smell at all rather than the "chemical" "fragrant" or "perfumey" scent of white smell.

These maps provide an interesting way to categorize and discuss smells, but the full extent of olfactory space remains unknown, ready for explorers to build more many-dimensioned maps of smell experiences. You can play around with the olfactory stimulus maps at the Sobel Lab website, look at the databases of rodent brain activity maps, and begin smelling the world more critically and enthusiastically today.