Thursday, October 27, 2022

Advances in Olfactory Perception

Scientists use machine learning to predict smells based on brain activity in worms
Jan 2022,

Putting this here because they used graph theory aka network science to decode the otherwise cacophony of neuronal crosstalk involved in smelling.

Also, why C. elegans? It has only 302 neurons, that's why:

Chalasani's team set out to study how C. elegans neurons react to smelling each of five different chemicals: benzaldehyde (almond), diacetyl (popcorn), isoamyl alcohol (banana), 2-nonanone (cheese), and sodium chloride (salt).

The researchers engineered C. elegans so that each of their 302 neurons contained a fluorescent sensor that would light up when the neuron was active. 

By looking at basic properties of the datasets—such as how many cells were active at each time point—Chalasani and his colleagues couldn't immediately differentiate between the different chemicals. So, they turned to a mathematical approach called graph theory, which analyzes the collective interactions between pairs of cells: When one cell is activated, how does the activity of other cells change in response?

The algorithm was able to learn to differentiate the neural response to salt and benzaldehyde but often confused the other three chemicals.

via Salk Institute, Cold Spring Harbor Laboratory and UC San Diego: Javier J. How et al, Neural network features distinguish chemosensory stimuli in Caenorhabditis elegans, PLOS Computational Biology (2021). DOI: 10.1371/journal.pcbi.1009591

a highly detailed, macro shot of a human nose, 8k, depth of field

The art of smell: Research suggests the brain processes smell both like a painting and a symphony
Apr 2022,

"These findings reveal a core principle of the nervous system," using a model to simulate the workings of the early olfactory system. This is a reminder that the olfactory system is an ideal model for understanding the brain.

In their computer simulation, they found that centrifugal fibers switched between two different modes -- one worked on a specific instant in time, while the other worked on the neural patterns across time.

This is where I make a further interpretation, which might be incorrect, but it seems like one is for comparing a smell to the body's repository (is this good or bad for me? have I smelled this before? where? who was I with?) and the other mode is for comparing the smell against itself, over time, perhaps to learn whether it's getting stronger or weaker. One uses autobiographical, physiological memory, and the other uses basic chemotaxis. One ontogeny and the other phylogeny?

Anyway, another reminder by one of the authors that the olfactory system is a good model: "Computational approaches inspired by the circuits of the brain such as this have the potential to improve the safety of self-driving cars, or help computer vision algorithms more accurately identify and classify objects in an image." -Krishnan Padmanabhan, associate professor of Neuroscience at University of Rochester School of Medicine and Dentistry

via University of Rochester Medical Center: Zhen Chen et al, Top-down feedback enables flexible coding strategies in the olfactory cortex, Cell Reports (2022). DOI: 10.1016/j.celrep.2022.110545

Sniffing out the brain's smelling power
Oct 2022,

(Out of order but seemingly related to the above) Here's another way of thinking of the two processes to smelling -- We said mitral cells are what do the smelling, but mostly because those were the ones we could see. Tufted cells were harder to see, up until now -- they find that the mitral cells were faster, more discriminating, and more broadly-tuned. 

The authors think the mitral cells only enhance important smells, but the tufted cells are part of a background process for identity and intensity. 

via Cold Spring Harbor Laboratory: Honggoo Chae et al, Long-range functional loops in the mouse olfactory system and their roles in computing odor identity, Neuron (2022). DOI: 10.1016/j.neuron.2022.09.005

a straight smooth vertical tube with the texture of human skin, highly realistic, hyper-real, 4k, Octane render 

Researchers map mouse olfactory glomeruli using state-of-the-art techniques
Apr 2022,

While other research teams previously examined the organization of glomeruli in the olfactory bulb, so far they only identified the positions of a limited subset of these clusters. As a result, the relationship between the location of glomeruli and odor discrimination has been very difficult to infer.

They used a combination of single-cell RNA sequencing, spatial transcriptomics and machine learning techniques. This allowed them to create a map that outlined the brain regions where most of the sensory neurons in the mouse olfactory bulb sent odor-related information.

via University of Massachusetts Medical School, Broad Institute of Harvard and MIT, and Stanford University: I-Hao Wang et al, Spatial transcriptomic reconstruction of the mouse olfactory glomerular map suggests principles of odor processing, Nature Neuroscience (2022). DOI: 10.1038/s41593-022-01030-8

How mosquito brains encode human odor so they can seek us out
May 2022,

Of the two nerve centers, one responds to many smells including human odor, essentially saying, "Hey, look, there's something interesting nearby you should check out," while the other responds only to humans. Having two may help the mosquitos home in on their targets, the researchers suggest.

First genetically engineer mosquitos whose brains lit up when active, and then deliver human-flavored air (with decanal and undecanal).

"When I first saw the brain activity, I couldn't believe it—just two glomeruli (out of 60) were involved. That contradicted everything we expected, so I repeated the experiment several times, with more humans, more animals. I just couldn't believe it. It's so simple."

via Princeton: Carolyn McBride, Mosquito brains encode unique features of human odour to drive host seeking, Nature (2022). DOI: 10.1038/s41586-022-04675-4

Thursday, October 20, 2022

Body Building for the Odor-Indulgent Robot

When Hidden Scents was written, back in 2015, it was certain that we wouldn't see an artificial smelling machine akin to the state of visual object recognition at the time, until we created robots that were indistinguishable from humans -- robots that grew up like a child did, complete with it's own physiological and emotional response to its environment, especially its social environment, and of course with its own resulting autobiography. 

Not until then would a robot, or even a simple computing device, be able to perform anything even close to the phenomenon of olfactory experience. And that's mostly because every one of us smells things differently, whether its genetics inherited before birth, anosmic dysfunction from viral infections, olfactory desensitization from old age, or just difference in cultural or personal preference. Without all this mess, no matter how hard you try to replicate the process of olfaction, you can't really call it the same thing as smelling unless you have an authentic adult human in the equation.

Granted there are reverse engineered electronic noses implanted on the bodies of roboticized locust cyborgs that have been programmed to detect explosives. Or less sci-fi e-noses used to detect fake whiskey. But that's not what we're talking about here. 

The headlines pasted below show us the dawn of the baby-bots, robots started from scratch, to understand themselves, and maybe even to accumulate an autobiographical identity. 

Researchers trained an AI model to 'think' like a baby, and it suddenly excelled
Jul 2022,

The exciting finding by Piloto and colleagues is that a deep-learning AI system modeled on what babies do, outperforms a system that begins with a blank slate and tries to learn based on experience alone.

via Princeton University: Luis S. Piloto et al, Intuitive physics learning in a deep-learning model inspired by developmental psychology, Nature Human Behaviour (2022). DOI: 10.1038/s41562-022-01394-8

Image credit: AI Art - Fragrance Advert by Magritte - 2022: portrait fragrance advertising campaign by magritte [link]

Engineers build a robot that learns to understand itself, rather than the world around it
Jul 2022,

Robot able to learn a model of its entire body from scratch, without any human assistance by creating a kinematic model of itself, and then using its self-model to plan motion, reach goals, and avoid obstacles in a variety of situations. It even automatically recognized and then compensated for damage to its body.

The researchers placed a robotic arm inside a circle of five streaming video cameras. The robot watched itself through the cameras as it undulated freely. After about three hours, the robot stopped. Its internal deep neural network had finished learning the relationship between the robot's motor actions and the volume it occupied in its environment.

Yeah I'm creeped out.

via Columbia University School of Engineering and Applied Science: Boyuan Chen, Fully body visual self-modeling of robot morphologies, Science Robotics (2022). DOI: 10.1126/scirobotics.abn1944.

Thursday, October 13, 2022

Headlines in the Smell World

Scientists thought they knew how the nose 'knows,' but new research suggests otherwise
Aug 2022,

I'm not doing a great job following this development, a reversal rather -- there's apparently some reconsideration that needs to be given to the way odorous molecules activate their respective olfactory receptors. Smell is the most understudied of all our senses, so it should be less of a surprise that one of the foundationary hypotheses of olfactory science needs some fine-tuning: 

G protein signal amplification is actually very low—so low that the probability of an odorant receptor activating just one G protein would be perhaps only 1 in 10,000. Yau said that, as such, the activation level "is very weak."

On a sidenote, what really stands out to me from this article is that the rhodopsin in the photosensitive cells on your retina (and all over your body in fact) are so sensitive they can detect a single photon of light. One single photon. And I thought our nose was sensitive. (It is, but for chemicals; the eye, and the photoreceptors in it, are for detecting the electromagnetic radiation beaming through our solar system.)

via Johns Hopkins University School of Medicine: Rong-Chang Li et al, Low signaling efficiency from receptor to effector in olfactory transduction: A quantified ligand-triggered GPCR pathway, Proceedings of the National Academy of Sciences (2022). DOI: 10.1073/pnas.2121225119

Friends at first sniff: People drawn to others who smell like them
Jun 2022,

The researchers found that the odor signatures of "click friends" were statistically more closely matched than odors between non-friends.
This study was done using the T-shirt test, an old trusty in smell science. 

I can't figure out which is more interesting here. The word "click friends" is pretty cool, never heard it. But this one is good: "Nonhuman terrestrial mammals constantly sniff themselves and each other and, based on this, decide who is friend or foe," wrote a group of researchers led by Inbal Ravreby at Weizmann Institute of Science in Israel.

"I don't sniff myself; I am not an animal; gross" as you completely obliviously brush your hair away from your face, or scratch your temple. The study that proved our absolute inability to avoid smelling our own hands (and the hands of everyone we meet, by way of our own hands that shook theirs) found that BEFORE the study even started, while people were still in the waiting room, they had their hand ready next to their nose 22% of the time! (see older post and the article itself). Filthy animals!

via Weizmann Institute of Science in Israel: Inbal Ravreby et al, There is chemistry in social chemistry, Science Advances (2022). DOI: 10.1126/sciadv.abn0154

People who consider olfaction important and actively sniff other's odors have stronger sexual desire
Aug 2022,

This study is based on questionnaires given to Chinese college students, and recall that there is a general understanding in the smell world that Asian people tend to not have the same scent-emitting glands as non-Asians; in other words, the deodorant market doesn't work very well in China. After their initial findings however, they started over and sent the same questionnaires to college students in both the U.S. and in India, and after all that:

  • Students who reported giving high value to olfaction or who actively sniffed other people also reported having a stronger sexual desire than others who responded.
  • Women tended to place more emphasis on smell than men, and reported lower levels of sexual desire in general.
  • Men in India reported stronger sexual desire than those in China and the U.S., and they also reported placing more importance on olfaction.  

A final thought -- culture in some ways can be a stronger mediator for olfactory perception than biology or genetics. Science like this is great, but it's only s very small piece of the full picture. I'm thinking of how the Marshmallow Test fell apart in a recent study because they considered that Japanese kids are conditioned to wait for everyone to be ready to eat at the dinnertable, and U.S. kids are not. 

via Southern Medical University in China and Technische Universität Dresden in Germany: Zi-lin Li et al, Sniffing of Body Odors and Individual Significance of Olfaction Are Associated with Sexual Desire: A Cross-Cultural Study in China, India, and the USA, Archives of Sexual Behavior (2022). DOI: 10.1007/s10508-022-02398-1

Rapid loss of smell predicts dementia and smaller brain areas linked to Alzheimer's
Jul 2022,
(But how about their sexual desire??)

via University of Chicago Medicine: Rapid olfactory decline during aging predicts dementia and GMV loss in AD brain regions, Alzheimer s & Dementia (2022). DOI: 10.1002/alz.12717

New study reveals where memory fragments are stored
Jul 2022,

This is about memory recall, the way we access memories stored in our brains, and it certainly does change the olfactory science, since we tend to consider olfaction as being deeply connected to our autobiographical memory via the hippocampus, which it is, but this suggests there are more olfactory details being stored in the prefrontal cortex than we thought:

While the overall experience is stored in the hippocampus, the brain structure long considered the seat of memory, the individual details are parsed and stored elsewhere, in the prefrontal cortex. This separation ensures that, in the future, exposure to any individual cue is sufficient to activate the prefrontal cortex, which then accesses the hippocampus for recall of the whole memory.

via Laboratory of Neural Dynamics and Cognition at Rockefeller University and Weill Cornell Medicine: Priyamvada Rajasethupathy, Prefrontal feature representations drive memory recall, Nature (2022). DOI: 10.1038/s41586-022-04936-2

Friday, October 7, 2022

Perceptual Cartography

This piece of news from from the Monell Chemical Senses Center, and it's supposed to be about how we can predict whether a molecule will have an odor or not. For example, the molecule for water, H2O, does not have a smell. Neither does a molecule of iron. Water smells, like all different kinds of things, and so does metal, but it's because of the fact that you can't find a place on Earth where life doesn't live, and life smells. So that includes how we deposit our skin flora on loose change and handrails, where select families of microbes proliferate in the microscopic textures and pores of the metal. Or how water can have any number of living (or dying) things in it, many of which can smell. And that's all because living things are walking, breathing chemical reactor factories that transforms molecules like their life depends on it, and these transformations give us single molecules of let's say isovaleric acid or cineole or delta-decalactone that offgas from the collective biosphere. 

But if we were to take any molecule, chosen at random, we don't have a good checklist to guess whether it will smell or not. And there's plenty of molecules still out there for us to discover, so how do we know where to start? That's where this study comes in. And for that purpose, it seems like it could be pretty helpful.

That's not what makes this article interesting to us here at Limbic Signal however. Instead, let's look at how this research tries to map out the information space of olfaction.

Information space is a hard concept to wrap your head around, because you're a meatbag bounded by the three dimensions of conventional reality (unless you're a robot reading this, of course). The entry point to discussions about perceptual information space start with vision. Is it light or dark, on a scale of one to ten, one being white and ten being black. That's one dimension where the space is a single line, and our "color" then exists somewhere along that line. 

Next, where does it fall on the rainbow, a spiraling spectrum that gradually changes from red to orange to yellow to green etc, each of which can be measured by it' radiating energy. Conveniently, the end can be shifted back to the beginning, so that after blue turns to violet, it keeps going, and we think we're approaching the end, but then violet starts to look like red, and we're back to the beginning again.

This gives us two dimensions, the light-dark and the rainbow (it's called hue but I'll call it rainbow here since it's a more familiar term. Color then has a light-dark number and a rainbow number, like an X-axis and a Y-axis, and any particular color exists at the intersection of these axes. This intersection is the information space of color. (There's typically an extra dimension used, called brightness or saturation, but let's just stop here.)

This information space allows us to group together similar colors, and to create different organizations of color combinations, based on color theory for example. And we can use charts that visualize this information space, so that we can communicate with each other about colors. Because as humans, we like to communicate about things. Some might say our ability to communicate, especially using verbal language, is what makes us human. 

But smells have a problem. There's no way to organize them, except to split them into "good" or "bad" categories, and since each of us has slightly and sometimes dramatically different opinions about whether a smell is good or bad, well, that distinction isn't very helpful.

Hot smells and cold smells? That doesn't mean anything. Wet and dry? (There's something here actually, related to chemical reactions that can and cannot take place underwater, because our sense of smell has been with us since we were fish, and now we're not, but we still hold some of that history.) It just doesn't work. Fruity, cheesy and burnt? Ok great, now what about the rest? And do they fit on a spectrum like the wavelengths of visible electromagnetic radiation? No they don't. 

Odor space is hard because instead of two dimensions like color (three really but who's counting), there are as many dimensions for smells as there are smells (and that's a lot). And that's just not very helpful if what we're trying to do is collapse the information, to make it easier to look at, think about, talk about...

And that's why this new research from Monell is so interesting, because it is one of the first comprehensive maps of odor space. It's huge, but it's useful because we've now managed to cross-off an even bigger world of molecules that simply can't be smelled.

What Makes a Molecule Smell?
Jul 2022, Monell Chemical Senses Center

Smell scientist Richard Gerkin produced a visualization of the odor space showing the universe of possible molecules and the regions of that universe where odorous molecules live.

A smell needs to be:
  • Volatile enough evaporate from the surface of its source
  • Not too volatile to still pass through the mucus layer coating the olfactory epithelium
  • Hydrophobic enough to escape the mucus layer and enter the binding pockets of olfactory receptors

via Monell Chemical Senses Center: Transport features predict if a molecule is odorous. Emily J. Mayhew et al. PNAS. April 4, 2022. 119 (15) e2116576119