People all over the world largely agree on what the colour blue looks like or what the shape of a ball feels like. But when it comes to describing odours, opinions often differ. This is because, unlike the processing of wavelengths of light in the brain, which makes it possible to determine colours relatively clearly, it is still not easy to deduce the smell of substances in our environment from their chemical composition. To help solve this so-called stimulus-percept problem, scientists at Friedrich Schiller University Jena, Germany have now presented data sets in which they compile how thousands of test subjects perceive, describe, and classify odours.
"The chemical structure of everything in our environment that we can smell is generally very complex. We are usually unable to say exactly what odour the individual chemical components emit, as the individual substances dock onto different receptors in the nose depending on their composition. For example, we don't know how a certain amount of carbon atoms smells," says Antonie Bierling from the Institute of Psychology at the University of Jena. In order to be able to make general predictions about how an odour affects people based on molecular properties, a great deal of information is required about how people perceive basic chemical building blocks.
On the way to an electronic nose
This is why scientists at the University of Jena have joined forces with colleagues at TU Dresden to create a fundamental odour database as part of the "Olfactorial Perceptronics" project. The project, which is supported by the Volkswagen Foundation, brings together various research disciplines: psychology, physics, chemistry, materials science, and medicine. For the database, they had over 1,200 test subjects smell 74 monomolecular – i.e. chemically very simply structured – odour samples. The test subjects then described what they perceived with their nose in their own words and also used a rating scale to assess, among other things, how pleasant or intense they found the respective odour. General statements about the odour of certain substances can be distilled from this information on perception. The researchers are also making their results available to the general public via an app ( https://crown-dataset.streamlit.app/ ).
In addition to the basic findings on odour perception, such databases pave the way for potential applications. "Our smartphone, for example, can recognize our face or our voice – but when it comes to digital smelling, developers are still coming up against fundamental limits," explains Alexander Croy, a physicist from the Institute of Physical Chemistry at the University of Jena. "With the help of such research results, however, we are already getting closer to the electronic nose and may even be able to identify our own body odour at some point."
Smelly feet in 13 languages
Such functions could have enormous significance for medicine, for example. The researchers in Jena have therefore joined forces with colleagues from Finland, Israel and the Czech Republic to develop another data set that records body odours. The joint project "Smart Electronic Olfaction for Body Odour Diagnostics" – SMELLODI for short – is funded by the European Union.
"We know that certain diseases have an impact on body odour. It can therefore be very helpful in recognizing and diagnosing illnesses to record it in detail," says Antonie Bierling. "However, this cannot be articulated well, as the vocabulary for describing body odour is still very limited." The researchers therefore asked over 2,600 test subjects in 17 countries online how they would describe the odour of individual parts of the body and how it differs when a person is ill or has been exercising.
This resulted in a catalogue of descriptions for various odours in 13 languages, which produces clear overlaps and thus allows general statements to be made about how certain areas of the body smell. The test subjects perceived armpit odour as sweaty, sour and stinky, they described bad breath as either fresh or stinky and foot odour as cheesy. Thanks to this broad database, scientists conducting research in this field now have access to a more standardized language system for describing odour perceptions. This database can also be accessed via an app (https://bow-descriptors.streamlit.app).
