Weizmann Institute of Science scientists have developed an "electronic nose" that is able to predict the pleasantness of smells

Possible future applications: neutralization of environmental toxins and digital transmission of odors

Weizmann Institute of Science scientists have succeeded in creating an "electronic nose" which is able to predict the degree of pleasantness that humans will experience when they smell unfamiliar smells. In doing so, the scientists contradicted the approach according to which the perception of smell depends on personal preferences and cultural contexts. In the study, which is published today in the scientific journal PLoS Computational Biology, the scientists show that the degree of pleasantness accompanying a certain smell depends on the molecular structure of the odorant, and that changes resulting from personal or cultural preferences only occur in certain contexts. These findings have practical implications in several areas, including bad odor monitoring, neutralization of environmental toxins, and odor scanning for the perfume and fragrance industry. In addition, they constitute another important step on the way to developing a method for the digital transmission of smells.

In recent decades, devices have been developed that serve as "electronic noses" capable of sensing odors and identifying them. The main component of these noses is an array of chemical receptors (sensors). When a particular odorant passes through the electronic nose, its molecular properties activate the receptors in a way that creates a unique electronic pattern - a kind of characteristic "fingerprint" for the odorant. Like a sniffer dog, the electronic nose must also be "trained" using smell samples, thus providing it with an initial database. The device will then be able to identify the odor samples presented to it by comparing their "fingerprints" with those found in the database. Unlike humans, when the electronic nose is presented with an unfamiliar smell, whose fingerprint has not been recorded and stored in the database, it will not be able to recognize or classify it.

The group of scientists from the Weizmann Institute of Science included Dr. Rafi Haddad - a research student (at the time) in the groups of Prof. Noam Sobel from the Department of Neurobiology and Prof. David Harel from the Department of Applied Mathematics and Computer Science, as well as Abba Madhani from the Department of Neurobiology at the Institute, and Dr. Yehuda Ruth from the Edith Wolfson Medical Center in Holon, used a unique approach to solve the problem. Instead of training the electronic nose by smelling certain odors, they taught it to determine the position of an odor on a continuum that rates the degree of pleasantness of the odor. In other words, they taught the electronic nose to predict whether the smell would be classified on the "pleasant" side of the continuum, its "unpleasant" side, or anywhere in between.

For this purpose, the scientists gathered a group of volunteers born in Israel, who were asked to rate the pleasantness of a selection of smells on a 30 unit scale: from "pleasant to a large extent" to "unpleasant to a large extent". Based on the data received, they developed a "pleasantness algorithm" which they fed into the electronic nose, and used it to predict the degree of pleasantness of a variety of new smells that were not included in the original database. The comparison of the rating made by the electronic nose, with the one made by a new group of volunteers, showed a high overlap of about 80%. When the scientists were "satisfied" with classifying the smells into one of two categories, "pleasant" or "unpleasant", the accuracy level of the electronic nose reached 99%.

To also test whether the perception of smell is personal and culture-dependent or general and independent, the scientists decided to test the predictions of the electronic nose against the rating of a group of volunteers who immigrated to Israel from Ethiopia. The results showed that even in this case the electronic nose reaches a rating similar to that received by the volunteers, even though its "programming" was done based on data collected from Israeli children. Although different smells have different meanings and contexts in different cultures, the predictions of the electronic nose reach similar achievements. This fact suggests that different cultures share a common basis in the perception of the pleasantness of smells.

Prof. Noam Sobel: "The fact that we can predict whether a person we don't know will like a particular scent, regardless of his cultural background, indicates that the pleasantness of smells is a basic biological property, and that certain aspects of the molecular structure of the substance determine whether its smell will be pleasant or not".

What, then, is the role of cultural influence?

"We believe that only in certain contexts, the cultural background does affect the perception of the pleasantness of smells. For example, many wonder how it is possible that the French love the smell of French cheeses, while the rest of the people think the smell is repulsive. In this case, we believe that the French do not think that it is a pleasant smell in itself, but that it indicates the quality of the cheese. If the smell was presented out of context - as a liquid sample in a jar - the French would also rate it as unpleasant, like any other person."

The scientists' findings - that the perception of smell is "embedded" in the molecular structure of the odorant, and their success in designing an electronic nose capable of classifying unfamiliar smells, could form the basis for new methods for scanning smells and environmental monitoring. In the future, it may be possible to use them as a basis for the digital transfer of smells - to accompany movies, games, etc. - to create a more tangible and complete sensory experience.