Sniffing out the mysteries of olfaction

More than two decades ago, a physicist walked into a bar in a Montana town. The physicist, Dmitry “Dima” Rinberg, was attending a neuroscience conference held at the town’s ski resort, and he found himself at something of a professional crossroads. He had a Ph.D. in the wave dynamics of superfluid helium but had realized he wanted to move into a field where he might break fresh ground. At the time, he was investigating sensory systems in cockroaches at the NEC Research Institute in Princeton, New Jersey.

In the bar, Rinberg struck up a conversation with a man named Alexei Koulakov, who was surprised to meet another physicist attending the conference. The men soon realized they also shared a similar life trajectory: They were both expatriates from the former Soviet Union and had grown up in Moscow, and both were looking to move beyond theoretical physics. By the time they finished skiing the next morning, they had become close, Rinberg says.

In the years that followed, both men shifted into studying the neuroscience of olfaction and became collaborators, with Rinberg building devices and performing experiments, and then developing theories with Koulakov. Even though it is thought that olfaction was one of the earliest senses to evolve, scientists know less about how we perceive smells than about vision or auditory processing. Joel Mainland, a member of the Monell Chemical Senses Center and adjunct associate professor of neuroscience at the University of Pennsylvania, notes that science has “not spent the same amount of time and resources on understanding olfaction” as it has on other senses.

Today Rinberg heads an olfactory research lab at New York University and leads a consortium at the National Institutes of Health called Cracking the Olfactory Code, which includes seven labs across the United States. Over the years, in collaboration with Koulakov and others, he has found evidence for what is called the primacy coding theory—a much-needed effort, given the hole in our knowledge of smell. But he has also leaned on his physics background to build scores of devices to deliver odor stimuli. These “olfactometers” are one of Rinberg’s “many contributions to the field,” Mainland says.

R

inberg was born into a Jewish family in Moscow in 1966 and remembers himself as a shy but happy child. In his early years he excelled at math, which his father supported by presenting him with textbooks. But despite his academic talent, a career in research was not guaranteed.

The Soviet government had a history of limiting access to higher education for Jewish students. For instance, in the 1970s and ’80s, at least three Soviet colleges were found to give difficult entrance exams to Jewish applicants, designed to effectively reject them. These colleges failed Jewish applicants on oral entrance exams by asking them so-called “killer questions” or “coffin problems”—mathematical queries that required extremely specific answers or were followed by questions of increasing difficulty. According to one report, the number of Jewish students in Soviet universities declined from 74,900 in 1935 to about 45,800 in 1960.

Against this backdrop, Rinberg’s mother explained to him when he was 9 that he would “have to work extremely hard to get anywhere,” he says. He heeded her advice and graduated in the top of his class before applying to Moscow State University’s Faculty of Mechanics and Mathematics. He thought he had done all he could, so it was even more painful when he did not get in. Instead, he pivoted to the Moscow Institute of Steel and Alloys (today the National University of Science and Technology MISIS), an institute with a constant need to train steel engineers and with fewer applicants, so Jewish students were accepted. There, he thrived.

During that time, Rinberg met his future wife, Tanya Tabachnik, while they were both preparing for a multi-week backpacking trip in the Tian Shan mountains in central Asia. Tabachnik was a serious student herself, studying electromechanical engineering at the Moscow Machine and Tool Institute, but even she was impressed by the enthusiasm and passion with which Rinberg talked about science. “I didn’t know people could love their studies so much,” she says. They married in 1989.

Soon after the Soviet Union began to fall apart, the country’s borders opened, and it became possible to study abroad. Looking for a place to earn a Ph.D., Rinberg applied and was accepted to the Weizmann Institute of Science in Israel, so the couple, who already had their first child, decided to emigrate. Yet their desire to leave was also partially due to the fact that Rinberg was a “free-spirited thinker,” says Tabachnik, and he wanted to flee the oppression of the Soviet Union. “We wanted our children to grow in a different place.”

He earned his Ph.D. in 1997. For his postdoctoral research, Rinberg found a position at the NEC Research Institute, studying cockroach escape behavior. Roaches sense changes in air flow as a mode of protection, and in 2000 Rinberg published a paper titled “Insect perception: Do cockroaches ‘know’ about fluid dynamics?” It was a crossover position to neuroscience, and he relocated his family to the United States.

I

n 2002, Rinberg moved to the Monell Chemical Senses Center in Philadelphia, where he studied olfaction. About a year later, Koulakov began working at Cold Spring Harbor Laboratory, studying neural development and brain evolution, and the two researchers developed a routine. Every so often, they met in Manhattan and spent the day walking around, drinking coffee, talking about their respective neuroscience research and sketching ideas in notepads.

At Monell, Rinberg turned fully to studying olfaction’s mysteries. The basics had been established: When an odorant molecule flies into a nostril, it binds to olfactory sensory neuron receptors via microscopic filaments called cilia. The neurons, in turn, send signals to the glomeruli on the surface of two onion-shaped olfactory bulbs. The glomeruli then relate the information to the brain, which ultimately perceives an odor.

Various odorant molecules bind to different receptor types. Humans have about 350 functional receptor types, dogs about 850, and mice more than 1,000. One theory posits that the loss of human olfactory receptor genes relates to the development and increased use of color vision, which made tracking smell less necessary for survival.

Yet exactly how we perceive smell was, and still is, a mystery. When photons hit our eyes, the red, green and blue color-sensing cones in our retina send signals to our brain, which determines the color we see. With hearing, we know that air pressure evokes auditory perception, and different frequencies and pressure strengths result in different pitches and loudness. But with olfaction, we don’t know the specifics of smell perception—how we smell coffee versus sour milk, for instance.

The lack of knowledge is partly due to the difficulty in understanding olfaction, Mainland says. With human color vision there are three types of cone photoreceptors, and with taste about 40, but with smell it’s almost an order of magnitude greater, what he calls an “exponential scaling in terms of complexity at the peripheral level.”

Another reason is that controlling smells in an experimental setting is extremely challenging. To study vision, a researcher can show a participant an image and then remove it. With hearing, a scientist can play a sound and turn it off. But once you expose a participant to a smell, there is no easy way to efficiently remove all molecules from the air.

Rinberg’s background in physics helped him tackle these problems. He gathered up lenses and cameras, and he connected brain electrodes and recording devices to capture neurons firing in response in real time. Tabachnik had tools-building training and sometimes lent a hand.

W

hile working at Monell, Rinberg stumbled upon an interesting phenomenon. During one experiment, he had recorded the activation of olfactory bulb neurons in a mouse in response to amyl acetate, which smells like bananas, while the mouse was under anesthesia—and he repeated the experiment when the animal was awake. Unexpectedly, when the mouse was asleep, the neurons responded more strongly than when it was awake.

Researchers commonly anesthetized mice to study their olfactory capabilities, but Rinberg’s results revealed this approach was problematic. “I was saying that the responses are very, very different from an awake state,” he explains. He analyzed the results with Koulakov, and they published the paper in the The Journal of Neuroscience in 2006. These findings surprised neuroscientists working in the olfactory field. “People didn’t like me,” Rinberg quips. But the paper also put Rinberg’s name on the olfactory map.

Shortly after, Rinberg joined the Janelia Research Campus, at the time a new center founded by the Howard Hughes Medical Institute, and his family relocated to the area a year later. In 2008, Janelia hired Tabachnik as its lead of tool development engineering.  A few years down the road, Rinberg started his own lab at New York University, where he continued to collaborate with Koulakov. Rinberg would gather data from his experiments and send them to Koulakov, who sometimes offered up hypotheses, Rinberg says.

In 2008, Rinberg began working with optogenetics, giving him the opportunity to interact with olfactory neurons directly. For instance, optogenetics enabled him to do experiments that circumvented the long-standing problem of lingering odorant molecules of smell. By shining light, he could stimulate olfactory neurons and interfere with a mouse’s ability to smell. Now Rinberg could investigate all sorts of things—such as how long it took a mouse to identify a smell, and how many neurons it had to activate to do so.

In 2012, he performed experiments on the length of time required for smell identification. His team used mice supplied by Northwestern University. The mice expressed a protein called channelrhodopsin-2 in their olfactory sensory neurons, which could then be stimulated by light. They also expressed a fluorescent protein in the olfactory receptor neurons of their nose and glomeruli, which changed its fluorescent intensity based on cell activity.

Finally, the animals had glass windows implanted in their skulls, and pressure sensors in their noses to time their sniffs. Then the researchers trained the mice to identify an odor A versus an odor B. If mice identified the odor correctly—and as a result licked the right spout—they were rewarded with water. At the level of neuronal activity, these actions looked like this: When the olfactory receptor neurons were activated by a smell, the signals travelled to the glomeruli on the surface of the olfactory bulbs, which lit up like Christmas trees.

Mice got the odors right on one sniff, which typically lasts about 300 milliseconds, but Rinberg wanted to know: Did they need all this time? So he designed more experiments that progressively shortened the amount of time mice were exposed to the scent by disrupting their recognition process with light. These further experiments established that mice could identify an odor correctly in less than half a sniff—or in about 100 milliseconds—after only the earliest and most sensitive receptors had been activated. Other receptors lit up after 100 milliseconds, but the mice seemed to be able to get all the information they needed “in one tenth of a second,” Rinberg says.

Once Rinberg and Koulakov had analyzed their data, they named this paradigm “primacy coding,” referring to the fact that only early receptors played a role in perceiving a scent. In 2017, they published the work in Nature Communications. “We don’t yet know how many [receptors] are necessary, but we’re arguing that only early ones really contribute to identification,” Rinberg says.

O

ptogenetics also enabled Rinberg to pursue another series of experiments to probe his theory: synthesizing odors and creating a perception of smell in a mouse’s brain. It’s almost like playing a piano, Rinberg says—the team can “play” a scent by lighting up a series of four receptors, for example, in a particular order. “We don’t know what odor it is,” Rinberg says, but the mice think they’re smelling something, and most likely it’s something they have never smelled before.

When the team changed the last-activated receptor in a series—number 4—the mice could still correctly identify a “synthetic” scent. If they changed receptor 2 or 3, however, the mice recognized the scent only 75 percent of the time. And if they changed the first receptor, recognition dropped to 50 percent, again proving the idea of primacy coding, Rinberg says. The team outlined their findings in a 2020 Science paper.

The receptors involved in early moments of smell recognition must be evolutionarily necessary, Rinberg and Koulakov hypothesized in their most recent paper, published in September 2024 in PLOS Computational Biology, because they have been used through generations. The researchers speculate that receptors not engaged in smell recognition became obsolete over time. “So if the receptor is not in the primacy group for any odors, we don’t need it,” Rinberg says, though he admits this is speculation and “very hard to prove.”

Not every olfactory scientist fully supports the primacy coding theory. Dale Matthew Wachowiak, a neuroscientist at the University of Utah School of Medicine who also studies the neurobiology of sensory systems, agrees that there’s “a lot of evidence” that the earliest responding receptors to an odor drive its perception but says many questions remain. For example, animals always take multiple sniffs of whatever they’re smelling; what do those additional sniffs tell them?

Wachowiak has also found that the first receptors responding to an odor can turn on in a different order. “[The activation series] can change,” he says, and the concentration of an odor seems to play a role too. A higher concentration can sometimes fire up less sensitive receptors, and they “actually come on first,” he says, meaning that the receptors that seem to be activated first are not necessarily the most sensitive to the odor.

Rinberg remains convinced of primacy coding, and these days he has a new goal: to build a “bionic nose” that can sniff out health problems before they might otherwise be diagnosed. He has created a startup around this idea, called Canaery, and says a bionic nose is “within the realm of possibilities.”

But he has an even more “sci-fi” aspiration. He would like to create a perception of not just any smell inside a mouse brain, but something specific—say, an orange, a lemon or cheese—a difficult task because it remains unclear which receptors register these smells.

If achieved, that accomplishment would open an entire new world of possibilities and would be “the first case when we create some object in the brain without presenting the object,” Rinberg says. “We can’t do it yet,” he adds, smiling. “But one day, I hope we will.”