Numerical ratings of odor character, an image of the odor vial, and an image suggesting "the hot tub is near"
Doty raises a point about the limits of our odor descriptors. Here's a vial from our lab. A trained panel described it as "Musty, Ozone, Earthy, Medicinal." A perfumer smelled it and said: "The hot tub is near." We have room to improve how we describe what we smell.
Doty replied, pushing back on several points. We see modern models predicting perception as half full and he sees them as half empty. doi.org/10.1093/chemse/bjag004
In 2023, machine learning models predicting odor character from molecular structure outperformed the median trained human panelist on a prospective test of 400 molecules. These models are now genuine tools, not speculation. doi.org/10.1126/scie...
The internet meme of a dress that looks blue and black to some viewers and white and gold to other viewers.
We argue that although odor perception is cognitively mediated, so is color vision. Color perception changes with brightness, genetics, culture, and context. That didn't stop colorimetry from becoming a precise and practical tool. doi.org/10.1093/chemse/bjag003
Doty reviewed the long history of classifying odors and linking them to chemistry. His conclusion: odor qualities are cognitive constructs. They don't map neatly onto physical dimensions the way color maps to wavelength, and computational models “haven’t fared well.” doi.org/10.1093/chemse/bjaf022
The odor limonene feeding into a neural network to predict the odor.
Can we predict what a molecule smells like from its structure? A new exchange in @ChemSenses debates this question. 🧪 @kingfunk.bsky.social
1/3 The humble glass jar is a workhorse in olfactory research, but comes w/ hidden problems:
- Unknown headspace concentration
- Concentration dilution by ambient air
These add noise to odor measurements.
Our preprint introduces something new, something bag.🧪 @jmainland.bsky.social & Matt Andres
High-Fidelity Tuning of Olfactory Mixture Distances in the Perceptual Space of Smell Through a Community Effort www.biorxiv.org/content/10.64898/2025.12...
Cool new post-doctoral position in my lab on the area of human social chemosignaling. What role does odor and olfaction play in shaping social behavior?
(Here also is a 10-minute video about this new line of work in the Human Nature Lab: humannaturelab.net/making-of/so... .)
Please share.
Scientists at Firmenich buff.ly/jq1TyYa and now Givaudan buff.ly/u2gP5Ow report several olfactory receptors that appear to map cleanly onto specific odor notes (earthy, grapefruit, woody, etc.). 🧪
New preprint from the very talented Bob Pellegrino! 🧪
Analysis of over 21,000 individuals identifies seven new genetic regions linked to odor perception, with some effects differing between men and women and connections to Alzheimer's risk. doi.org/g9vn4j
In May we launched the DREAM Olfactory Mixtures Prediction Challenge 2025. The leaderboard round closed yesterday and 71 teams submitted predictions! Even if you missed the leaderboard phase, you can still submit a final model for the August 8th deadline.
My favorite part of these challenges is actually after the competition closes. All participants get together and work collaboratively on a manuscript summarizing key insights and where the field stands. Join us! www.synapse.org/Synapse:syn6...
Building on last year's Mixture Discriminability Challenge, we're offering a dataset of 650+ blends, each assessed by trained panelists using a 51-word scent lexicon.
Most existing models predict odor perception at just one concentration. Yet, some odorants dramatically shift in quality as concentration changes—such as grapefruit mercaptan, pleasantly fruity at low concentrations but distinctly sulfurous at high levels.
This year, we have two challenges: 1) Predicting how odor quality changes with concentration, and 2) Describing an olfactory mixture. You can make up to three leaderboard submissions until July 21st, with one final submission allowed before August 8th.
Last year, we ran the second DREAM Olfaction Prediction Challenge www.synapse.org/Synapse:syn5... . This time focusing on predicting the similarity of mixture pairs.
Ten years ago the first DREAM Olfaction Prediction Challenge launched to predict the smell of a molecule from its structure. This led to a publication establishing benchmarks doi.org/10.1126/scie... , and then a subsequent surge in predictive modeling.
With a contemporary smartphone, capturing, transmitting, and replicating sounds and visuals is a breeze, yet replicating scents remains elusive. If you want to change this, consider joining the 2025 DREAM Olfaction Challenge at www.synapse.org/Synapse:syn6....
@oldbap.bsky.social
After 21 years at my dream job, I’m very sad to announce my early retirement from the National Institutes of Health. My life’s work has been to scientifically study how our food environment affects what we eat, and how what we eat affects our physiology.
Check out our review comparing typical experimental to naturally-occurring odor concentrations. You might be surprised.
www.jneurosci.org/content/45/1...
Thanks to Matt and Betty for asking me to take part!
A long post about what’s happening to the science funding agencies in the US and why. As mentioned, this one just kept getting longer even as I kept stripping curse words from it.
www.science.org/content/blog...
SOME GOOD NEWS ABOUT NIH FUNDING
Closed sessions of institute and center Advisory Counsel meetings are now permitted. This is where grant applications are discussed and approved for potential funding.
Open sessions where there are public presentations and discussions are still not allowed.
Alexander Fjaeldstad and I have a new chapter in the book Smell, Taste, Eat: The Role of the Chemical Senses in Eating Behaviour, edited by @LorenzoDStafford. Lots of great insights here.
From our chapter, it’s clear that smell loss and its impact on ingestion remain vastly underexplored."
doi.org/10.1073/pnas...
Olfaction in spiders. We found that the wall pore sensilla in spiders look similar from the outside to those of insects, but differ ultrastructurally in various ways. Independent evolution!
🧪
Our DeepNose paper is in arxiv. In it, we present an equivariant CNN which can predict human olfactory perception based on molecular shapes, including mixtures. Comments/suggestions are very welcome.
arxiv.org/abs/2412.08747
A background in physics, and his own curiosity, have helped Dmitry Rinberg tackle the complexities of the neuroscience of smell. More on Rinberg sniffing out the mysteries of olfaction.
By Lina Zeldovich
www.thetransmitter.org/olfaction/sn...
The core figure, showing review counts from Publons data, indeed shows the intended effect but buries the lede:
No one uses Publons anymore.