Thursday, May 20, 2021

Olfactory Overload

For centuries, smell has been considered the lowest sense. Even Science itself has avoided it. No more; olfaction is experiencing an absolute revolution.

Oh really? Yes. Consider for a moment, the €2.8M price tag on this European olfactory heritage and sensory mining project, called Odeuopa. That's a nice price tag for research on the history of smells.

I review a few hundred articles every week, pulling aside anything I see on olfaction. Some weeks there's a good article, maybe two, sometimes nothing. Much of the time, the article is about a development in hi-tech sensors or electronic noses, or how the East doesn't like "new car smell" and what Western car companies are doing about it... . 

But now, about once a month for seven months in a row, another piece of research surfaces to enrich our understanding of olfactory perception. Not just hype-cycle blurbles, but domain-foundation scientific breakthroughs are redefining how we think about olfaction, and our models of how it works.

The news is that each one of us employs a dynamic, combinatorial chemosensory system. It's capable of adapting in real time to an ever-changing environment, and it's using a combinatorial network of hundreds of genetically-determined olfactory receptors working in unison to identify and interpret any possible combination of odorous chemicals that we could ever be exposed to. 
First, a reminder of what it means to be using a "combinatorial" approach to perception. Epistemologically, combinatorics comes from mathematics, but combinatoric optimization and combinatorial dynamical systems are subfields of this domain, usually found in areas like graph theory or network theory. These areas overlap with the "brains" of our early 21st century artificial intelligence machines -- the deep learning neural networks you should be hearing about daily. 

But what does it mean for olfaction to be combinatorial? It means that olfaction is all gestalt. We don't use one type of neuron to smell one type of smell. We use a bunch for each, and they overlap too. In other words, it's a mess.

Some odorants, in theory, could activate (or inhibit) every receptor we have (roughly 400 functional). And it's the combination of all those excitations and inhibitions that create odor identity. That's a lot of combinations. And you would need all of them to identify that one odorant. And if you lost only one, by viral infection for example, that thing would not smell the same. In reality, this is not how it works because it's a lot more complicated, and there are so many exceptions to the rule that it's barely a rule. But it's getting clearer by the day. 

The main point of a combinatorial system is that you can't "map" it (it's a mess, remember?). This is something Science has been trying to do for a long time. Using language as an intermediary, this attempt to map the olfactory dimension started with the Dravnieks dataset, a bunch of odorous molecules mapped to descriptors produced by people who smell those molecules:

Dravnieks A. Atlas of odor character profiles. Philadelphia: ASTM; 1985.

Arctander is also used to organize the aromasphere by way of language: 

Arctander S. Perfume and flavor chemicals (aroma chemicals). Montclair, NJ: Author; 1969.

But then things changed. The DREAM dataset is produced, using huge chemoinformatics datasets for the individual molecules, mapped against equally huge semantic analysis datasets made of a bustling lexicon of odor words. This paper via Leslie Vosshall's lab in Rockefeller University sums it up:

Keller A, Vosshall LB. Olfactory perception of chemically diverse molecules. BMC Neurosci. 2016 Aug 8; 17(1):55.

Multidimensional Folds - Lach - 2010

And now onto 2020, a big year for olfaction. Figures so much of this will be overshadowed by a raging global pandemic.

In March, a study from Hebrew University shows us that our olfactory receptors are not just activated but inhibited, and that they can change over time from one to the other. Back in 2006, Wilson and Stevenson's book Learning to Smell investigated this idea of the blank slate. The thing is, optogenetics wasn't invented yet. Well it may have been invented, but they weren't slipping glass fibers into mouse neurons to monitor their activity in real time. We're not looking at the psychology of smell anymore but the actual neurological behaviors of it. 

And they see that the receptors themselves do in fact learn, change, adapt, and even revert back to a previous state. I've repasted this description already elsewhere on this site, because it's such a big deal, but again:
The general profile of excitatory vs. inhibitory responses by mitral cells changed with learning and task demands. In naive animals, most responsive cells (71%) responded by excitation to the odours. Following the learning of the 5-decision boundary task, the ratio of excitatory/inhibitory responses reversed. After learning, the majority (71.4%) of neurons now responded by inhibitory calcium transients to the odours (Fig. 6C,E). The ratio of inhibitory vs. excitatory responses reverted back to normal after retraining the mice on the 1-decision boundary task. Specifically, 73.3% of responsive neurons were again excitatory on day 18.
via Hebrew University: Flexible Representations of Odour Categories in the Mouse Olfactory Bulb. Elena Kudryavitskaya, Eran Marom, David Pash, Adi Mizrahi. Hebrew University of Jerusalem. Mar 24 2020. BioRxiv. doi:

April. When you hear the phrase "more than the sum of its parts," that's codeword for things combinatorial. This one doesn't use the optogenetics described above, but an imaging technique called SCAPE microscopy:

Making sense of scents - 3-D videos reveal how the nose detects odor combinations
Apr 2020,
Using a cutting-edge 3-D imaging method called SCAPE microscopy, the Columbia team monitored how thousands of different cells in the nose of a mouse responded to different odors—and mixtures of those odors. They found that the information that the nose sends to the brain about a mixture of scents is more than just the sum of its parts.
The researchers expected to see that the cells activated by mixtures of odors would be equivalent to adding together responses to individual odors. In fact, they found that in some cases an odor can actually turn off a cell's response to another odor in a mixture [previously known via violet ionones]; in other cases, a first odor could amplify a cell's response to a second odor.
The team's data challenged the traditional view that the brain makes sense of a mixture of scents by figuring out all of the individual components. It confirmed what perfumers have long known: combining different scents can create a certain experience on its own, essentially becoming an entirely new scent that can provide a completely different experience.
via Stuart Firestein's lab at Columbia University: L. Xu el al., "Widespread receptor-driven modulation in peripheral olfactory coding," Science (2020).

Multidimensional - Oliver Panthsdown - 2008

Skipping May, June shows us "synthetic olfactory perception" which is exactly what it sounds like.

Researchers at New York University's Langone Health Center simulated olfactory perception with a synthetic electronic odor signal. In laymen's terms, mouse noses were tricked into thinking they smelled something when it was actually just an electrical signal. This is kind of like the way you can open someone's skull and zap certain parts of their brain, and they will feel tingles in corresponding parts of their body, even though you're not touching those parts of their body (don't try this at home though).

There are also some interesting results from this study that support the mostly-uncontroversial yet definitely misunderstood theory of information processing in the olfactory bulb, which is that the detection of odor-representations is more of a combinatorial process, and less of a one-to-one system of odor molecules and neuron receptors. And, this combinatorial perception theory is a primary reason as to why we cannot comprehensively organize olfactory experience into subsets or primary odors. (And the reason for writing a book about the language of smell.)

via NYU Langone: Manipulating synthetic optogenetic odors reveals the coding logic of olfactory perception. Edmund Chong, Christopher Wilson, Shy Shoham, Stefano Panzeri, Dmitry Rinberg. Science 19, Jun 2020, Vol. 368, Issue 6497, eaba2357. DOI: 10.1126/science.aba2357

Later on, in July, Harvard Medical School releases a similar study, showing "flexible cortical representations in odor space" here:

via Harvard Medical School: Stan L. Pashkovski et al, Structure and flexibility in cortical representations of odor space, Nature (2020). DOI: 10.1038/s41586-020-2451-1. [alt link]

Sum-of-its-parts strikes again, July. 

Engineering and philosophy combine for an emerging understanding of smell
Jul 2020,
Shi Nung Ching of the Preston M. Green Department of Electrical & Systems Engineering, and doctoral student Sruti Mallik developed computational models of neural circuits that mimic the sensory act of smelling. They found the models also manifest certain properties analogous to those observed in olfactory sensory processing in insect brains.

Researchers found that their sensory system model developed emergent properties—properties that are more than the sum of their parts, so to speak—similar to properties seen in an insect's antennal lobe, which is important for its sense of smell.
via Washington University in St Louis: Sruti Mallik et al. Neural Circuit Dynamics for Sensory Detection, The Journal of Neuroscience (2020). DOI: 10.1523/JNEUROSCI.2185-19.2020

Astral Fragments - Stacy Young - 2016

Onto August. They're looking at the rat hippocampus; because that's the place where memories are stored, and it's hardwired into the olfactory system, the limbic system. They're showing that the brain identifies things differently over time. In a way, saying that there is no objective reality, only subjective multitudes:

via the University of Technology Sydney: Laura A. Bradfield et al. Goal-directed actions transiently depend on dorsal hippocampus, Nature Neuroscience (2020). DOI: 10.1038/s41593-020-0693-8

September, "sum of parts" again:

Nose's response to odors more than just a simple sum of parts
Sep 2020,
"New research from Kyushu University shows that a much more complex process is occurring, with some responses being enhanced and others inhibited depending on the odors present."
via Kyushu University: Shigenori Inagaki et al, Widespread Inhibition, Antagonism, and Synergy in Mouse Olfactory Sensory Neurons In Vivo, Cell Reports (2020). DOI: 10.1016

October shows us that you can teach yourself to smell better. I am compelled to remind the reader that olfactory receptor cells are the only part of your brain that pokes outside the body, making them very vulnerable. This also makes them a great point of entry for viruses invading the body, but it's also the reason why these cells regenerate profusely throughout most of our lives. And that's a reason why you can train yourself to smell better. 

These scientists basically stopped sending odors in the air to one nostril, and found that neurogenesis slowed down (use it or lose it). The idea is that as these cells re-grow, they may be changing the cell types in order to adapt to a changing environment. This is called stimulation-dependent neurogenesis, and although it's still up in the air as to how it all works, get in on the ground floor:

Study finds odor-sensing neuron regeneration process is adaptive
Oct 2020,

via University of Colorado Anschutz Medical Campus: Carl J. van der Linden et al, Olfactory Stimulation Regulates the Birth of Neurons That Express Specific Odorant Receptors, Cell Reports (2020). DOI: 10.1016/j.celrep.2020.108210

November now. Not olfaction specifically, but memory, which is closely related. The common theory has been that each memory gets its own neuron, but now an alternative model is ascending, and it looks more like the same group of neurons store all memories.

All the data we have on this stuff comes from fMRI. But fMRI can't see individual neurons. If we look at the neurons one at a time, we see something very different happening. 

This is certainly an idea to get familiar with. It should also fill you with wonder at what else we will figure out with rapidly-advancing neuro-tech:

Human intelligence just got less mysterious, neuroscientist says
Nov 2020,

via the University of Leicester: Rey HG, Gori B, Chaure FJ, Collavini S, Blenkmann AO, Seoane P, Seoane E, Kochen S, Quian Quiroga R. Single Neuron Coding of Identity in the Human Hippocampal Formation. Current Biology : Cb. PMID 32142694 DOI: 10.1016/j.cub.2020.01.035 

No comments:

Post a Comment