Scents are challenging to predict, but new research may change that. Image Source: Flickr User: vasile23
Human perception of physical sensations and inputs is generally predictable when it comes to colors and sounds, which can be predicted based on wavelength. But scents aren’t quite as straightforward, so new crowdsourced research has set out to delve nearer to the root cause of human perception of scent.
Researchers had volunteers smell a diverse 476 molecules and interpret the scents based on how strong they found it, if it was pleasant, and whether it evoked nineteen different attributes, such as “flower,” “sweaty,” and “bakery.”1 The results were then assessed for links between molecular structure and interpretation of scent. Links began to appear, commonalities that could be applied to CPGs with innovative lab software.
A Rose By Any Other Name
Scent is a strange thing to most people, as its interpretation is seldom agreed upon. Human perception of scent is attributed to the sets of olfactory receptors that line the respiratory tract. There is a great degree of variability in this portion of human DNA, and its roots are likely attributed to primal survival mechanisms. That said, nowadays, you may have the gene/allele for certain olfactory receptors whereas your neighbor may not. Additionally, you may interpret one scent as being “cold,” whereas your friend may smell something “acidic.” This may sound like a banal distinction, but it helps explain how many people perceive cilantro as tasting like soap. Although humans have mostly evolved beyond relying on smells or tastes to interpret food as poisonous, it is important when looking to sell CPGs that rely on taste or smell.
Some of the most easily predicted scents from the models that were generated based on the data were “garlic,” “fish,” “pleasantness,” “sweet,” “fruit” and “spices,” but this may primarily be due to consensus amongst participants with regards to how these smell. Other attributes, such as “warm,” “wood” and “acidic,” were a bit more challenging to predict. The researchers saw some correlation between easily definable attributes and familiarity. The researchers drilled down to assess the molecular features at the root of these attributes. With one of the models, it appears that 15 molecular features allowed researchers to reach 80% predictability. These features can be used in silico with modern lab software to enhance the perfuming process in many CPGs, saving both time and R&D dollars. This new predictive modelling of scent is still in it’s early stages, but there is a lot of potential to be unpacked, and streamlining current data may save a substantial amount of time and money.
Designing Prolific Scents
There is a certain degree of safety in designing CPGs with familiar scents. They are generally a safe bet, but there is room for tweaking and forging new paths:
- Specific Tweaking of Familiar Scents. During a season change, particularly the transitions to fall and winter, familiar scents play a starring role. Some common CPG manifestations of this are soap and scented candles. Pumpkin pie or pine-scented candles are particularly common around Thanksgiving and Christmas, but by uniting recent research into scent perception with innovative lab software, it may be possible to more easily tailor scents to be more specific. Say your company was producing candles and your consumer research data indicated that shoppers in a particular region preferred a crisper scent in their dessert-scented Thanksgiving candle. Rather than working through hundreds of trial scent runs, R&D time could be preserved through modelling these scents using innovative lab software, refining them, predicting their compatibility with your candle base and then conducting a limited with fewer scents.
- Creating Original Scents – Signature scents are a way to define a product or a brand. Chanel No. 5’s unique scent has contributed to the staying power of that perfume over a number of generations. Consumer research may indicate that your target audience has a very particular set of scent attributes in mind for your CPG, which may allow to stray beyond familiar scents and create your own. By modelling molecules in silico using modern lab software, your R&D department can create and predict a unique scent that can define your product.
With BIOVIA Materials Studio, comprehensive models can be created to allow researchers to better predict scents based on molecular features and assess how they will integrate with the materials in your CPG. This will save time and money spent on R&D and may allow you to hit a broader audience or set your product aside as unique and coveted. Please contact us today to learn more about how our software options can support the efforts of your lab.
- “Predicting human olfactory perception from chemical features of odor molecules,” February 24, 2017, http://science.sciencemag.org/content/355/6327/820.full ↩