• Odorbet

Installation 26: Digiscents & Noble Prize in Olfaction

Updated: a day ago

Collection of Historical Scientific Instruments (1997-1-1960)–ca 1938 (Image osmoscope)

“You don’t want to just do a problem because it’s easy to solve, you want to do something that you’re obsessed with, that you just have to understand, because that’s where the joy comes from, and that also, I think, is where the great discoveries come from, for people are really trying to try to figure out things that they don’t understand.” - Linda Brown Buck(1)

Scent remains the most stubbornly analog of all the art mediums. The only thing you can entirely digitize on the olfactory front is the marketing message. There are specialized AI devices, VR devices with scent and some promising experiments to date, but nothing truly affecting the consumer or creative maker quite yet. Last month I attended a talk by Alex Wiltschenko (Google Brain, head of Machine Learning for Olfactory Learning), where I learned that machine learning is responsible for some incredible things to date that work for the private and public sectors in both harmonious and helpful ways and some less savory ways. Computers can now speak and hear, and thanks to Alex’s group they will likely begin to smell and be smelled.

I’m simplifying for sake of a blog entry, suffice to say that they are hoping to get a reliable model for predicting odors based on a chemical’s (scent’s) molecular structure. This odor space model would then be the basis for a possible osmoscope, or digital camera for smell. The structure-odor relation problem(2) stands in the way of a predictive modeling system, though they continue to press on trying to figure out the organizational principal of odor space.

Another group a few years ago also was researching how to build such a predictive model based on data collected. This was called the Olfaction Prediction Challenge and was organized by Pablo Meyer with data collected by Andreas Keller and Leslie Vosshall.

Artists and fragrance houses remain ahead of the game from a creative standpoint where their intuitive, analog(3), and tightly held mixture data sets reign supreme. Alex noted that Google needs to team up with this information in order to move the predictive model along. In other words the mixture data set (versus the single note) will yield more applicable and describable smells, so the computers will actually be able to learn more. Right now the machine learning project for scent is stunted to a degree without access to these mixture datasets.

If Google does crack this conundrum, the code for an osmoscope would be under lock and key just as their Google Translate code is proprietary. In the meanwhile I’m going to think about the days ahead where we push notes to one another on our digital devices that say “I wish you were here!” that is coupled with a scentogram of where we are and what we are doing. Or perhaps we can choose from a library of our own smells on our device and send accordingly. It feels fun and mad at the same time.

Sending a poof of digital lemongrass and lilac to kick off your summers!


Linda Brown Buck (born 1947) is an American biologist best known for her work on the olfactory system. She was awarded the 2004 Nobel Prize in Physiology or Medicine, along with Richard Axel, for work on olfactory receptors. In 1991 Buck became an assistant professor of neurobiology at Harvard University where she expanded her knowledge of the nervous system.

AND……..Buck was awarded the Takasago Award for Research in Olfaction (1992), Unilever Science Award (1996), R.H. Wright Award in Olfactory Research (1996), Lewis S. Rosenstiel Award for Distinguished Work in Basic Medical Research (1996), Perl/UNC Neuroscience Prize (2002), and Gairdner Foundation International Award (2003). In 2005, she received the Golden Plate Award of the American Academy of Achievement. Buck was inducted into the National Academy of Sciences in 2003 and the Institutes of Medicine in 2006.[18] Buck has been a Fellow of the American Association for the Advancement of Science and the American Academy of Arts and Sciences since 2008.She also sits on the Selection Committee for Life Science and Medicine which chooses winners of the Shaw Prize. In 2015, Buck was awarded an honorary doctorate by Harvard University and elected a Foreign Member of the Royal Society (ForMemRS).

Osmoscope - 1. An instrument for measuring odors, originally invented in 1938 to measure the intensity of odors by Gordon Fair and William Wells. 2. A digital camera for capturing and sharing scent.

Scentogram - The equivalent of a visual jpeg file, where it is a compressed scent file for sending accross multiple platforms. (Completely made up - Catherine Haley Epstein, also the name of a font designed by Jakob Fischer in 2008 per a google search)

(2) Structure-odor relations

“Structure -odor relations (SOR), tell us about the mechanisms of human olfaction and about the synthetic chemistry of odorants. A perennial difficulty of structure-odor relations has been that both structure and odor have proved hard to pin down. Considered as a structure-activity problem, olfaction is several orders of magnitude more complicated than its conventional pharmacological counterparts because there are many more structures and a vast number of odors. There is also an additional problem: as a sensation, olfaction does not seem to enjoy the same status as, say, vision. Most biologists, indeed most people not directly involved with fragrances or flavors seem to think that odor sensation is "subjective" and not necessarily shared by others. It is striking how few experiments in which odorants are applied to biological preparations take into account the perceived odor of the molecules. We hope that biologists will realize that, once a vocabulary is agreed upon, odor is as reliable a sensation as pitch or color.” (Turin, L. Yoshii, F., 2003/01/01 Structure-odor relations: A modern perspective, Handbook of Olfaction and Gustation

(3) Realizing there are many spreadsheets involved in fragrance creation these days.

Recent Posts

See All