How Smell Shapes Our Hormones, Behavior and Friendships, According to Dr. Noam Sobel

Andrew Huberman opens this Essentials episode, which revisits past episodes for science-based tools for mental health, physical health and performance. He introduces Dr. Noam Sobel as a neurobiologist at the Weizmann Institute of Science and an expert on smell.

Huberman says he wants the conversation to cover chemosensation, especially olfaction, and related perceptual or subconscious abilities such as pheromones. He begins by asking Sobel to outline the components of the human sense of smell.

Sobel explains that primary olfaction, which he calls orthonasal olfaction, works through the nose as people sniff. Airborne molecules travel about 6 or 7 centimeters up the nose to the olfactory epithelium, where they interact with receptors.

Sobel says humans have about 7 million such receptors, which may be in roughly 350 kinds. He notes that a meaningful percentage of the genome is devoted to olfactory receptor types. When odorants dock at a receptor, they are transduced into a neural signal in the form of action potentials, which travel along the olfactory nerve through the perforated thinnest part of the skull.

The nerve synapses at the first target in the brain, the olfactory bulb. Sobel describes the pattern of this synapsing as the most extreme case of convergence in the mammalian nervous system: all receptors of one subtype converge to one location in the bulb. He also gives an example in which coffee activates receptor types 1, 3 and 7, so the glomeruli for those receptors are activated.

Sobel says humans are utterly remarkable in their sense of smell, though this point often needs repeating for audiences outside olfaction research. He contrasts human behavior with dogs, which visibly sniff each other when they meet. Humans do not typically walk up to a stranger and sniff them carefully, although people are expected to sniff their babies and it is not taboo to sniff loved ones.

Sobel says his lab is finding more mechanisms in which humans do sniff others, and he gives handshaking as an example. He notes that the origin of handshaking is poorly understood, and that the common explanation, that it shows one is not holding a weapon, has no good evidence behind it.

The lab started noticing that people often touched their faces after shaking hands.

Huberman notes, for listeners, that Sobel was describing people wiping their hands on their faces after a handshake. Sobel says these behaviors are common and easy to overlook. He describes a study led by Idan Frumin in his lab to test whether the behavior is real.

Participants were brought into an experiment room and told the study would start soon, without knowing its purpose. Unbeknownst to them, they were being videoed while they waited three minutes. The researchers later measured their baseline face-touching, which Sobel says was not low. A new experimenter then came in, and in half the cases offered a handshake.

Sobel says the results showed that people who shook hands touched their noses more and sniffed their hands, which confirmed the behavior was olfactory. People also smelled the other person's hand, not only the one that shook hands. Sobel says the lab thinks this points to a comparison of self and other.

Huberman takes a break to acknowledge Element, an electrolyte drink with sodium, magnesium and potassium and no sugar. He says that proper hydration matters for brain and body function and that even slight dehydration can reduce cognitive and physical performance. He says he dissolves one packet in 16 to 32 ounces of water in the morning and during exercise, and he mentions the flavors he likes, including a limited-edition lemonade. Listeners can claim a free sample pack with a purchase at drinkelement.com/huberman.

Sobel describes a study published by Inbal Raviv in his lab on click friendships, which are relationships formed right away. Raviv's theory was that similarity in body odor may contribute to such clicks, so that people who smell alike may click with each other. Other explanations, such as looking alike or sharing interests, could also be possible.

To test this, Raviv recruited click friends across Israel through social media. The participants were same-sex, non-romantic pairs of friends, men and women, whose friendship started as a click. Each participant was given odorless shampoo and soap to use for three weeks. Then they slept two nights alone in the same T-shirt, from which the body odor was extracted.

Sobel says the lab has a method for extracting body odor from T-shirts. Raviv first asked whether click friends are more similar in body odor than would be expected by chance. She tested this with an electronic nose, which Sobel describes as a poor effort to mimic the mammalian nose: a set of sensors that respond to airborne molecules, in this case metal-oxide sensors.

She found that click friends were indeed more similar to each other than random dyads, and Sobel calls this a significant difference.

After the electronic nose results, Raviv had other participants smell click friends and non-click friends. They judged the click friends as more similar to each other. Because click friends might share environments that make them smell alike, Raviv also tested total strangers.

The strangers were first smelled with the electronic nose and then engaged in a completely nonverbal social interaction called the mirror game, in which one person moves their hands and the other follows. One prediction, that similar-smelling people would follow each other better in this game, did not hold up. In a round-robin in which everyone played with everyone, participants then rated how much they would want to be friends with each other and how nice and affectionate the other person seemed.

Sobel says all of these ratings were predicted by the electronic nose. People who smelled more similar were more likely to think the other person would be a friend and a nice person. He says this shows smell can predict friendship and plays a causal role in building friendships, rather than merely reflecting them.

Huberman asks about other so-called pheromone effects and chemosensory signaling between individuals. Sobel says much of this becomes a question of semantics. In the lab's publications, they do not use the term pheromone, because it would not help and would probably only hurt the work. Instead, they talk about chemosignals, and Sobel says humans clearly emit chemosignals from their bodies that influence other people's behavior.

Sobel says the flavor of the month in the field for the past decade has been what is called the smell of fear. He says humans emit a particular body odor when they are in fear, and this body odor influences other people by raising their autonomic arousal and sympathetic state. In effect, fear may be somewhat contagious.

Sobel says people have long said dogs can smell fear in humans, but this was only effectively shown in a study about a year and a half earlier. Humans can also smell human fear. Several labs, starting with Denise Chen and Haviland Jones and including Sobel's own lab, collected body odor from people when they were in a state of fear and from the same people when they were not. Other people could determine which state the odor came from, and this influenced their behavior.

Huberman recalls a guest on his podcast, Charles Zuker, a Columbia professor who has done work on vision, olfaction and thirst sensing. Huberman says Zuker described something in the breath of romantic partners that is hopefully appetitive rather than aversive. Zuker also talked about the smell at the nape of his grandchildren's necks, and how he misses that smell when he thinks about missing them, because that smell is associated with the feeling.

Sobel says graduate student Eva Mishor was interested in aggression and initially set out to study chemosignals of aggression, including collecting body odors at MMA clubs. The work never went anywhere, Sobel says. At the same time, Heinz Breer was studying hexadecanal, a molecule that acts as a chemosignal in mice to promote social buffering, meaning mice feel good when they are together in a group. Mice emit it in their feces.

Breer and a colleague discovered its receptor, which is highly conserved across mammalian evolution, and they hypothesized it might be a universal mammalian signal. Because the lab worked on humans, Breer asked them to study it in people, and eventually he sent them hexadecanal. Mishor then used sweat from human participants and built a standard paradigm to study human aggression.

Mishor used the Taylor Aggression Paradigm, a well-validated standard protocol that quantifies aggression. Participants played it either under exposure to hexadecanal or in a control condition. In men, hexadecanal consistently reduced aggression, and the effect size was meaningful and unusually strong compared with aggression research generally.

In women, however, hexadecanal increased aggression to an equally significant degree. Sobel says this made no sense to him at first. Mishor repeated the experiment inside an fMRI magnet to track brain activity, and the results replicated: hexadecanal made men less aggressive and women more aggressive.

Sobel says Mishor's insight was that the pattern makes sense from the perspective of a mammalian offspring. Paternal aggression is often directed at offspring, including infanticide, and sadly male aggression toward human children also occurs. Maternal aggression is often protective.

If an offspring has a molecule that makes its mother more aggressive and its father less aggressive, both effects are beneficial to it. Sobel notes that hexadecanal is the most abundant semi-volatile coming from baby heads. He says babies may be conducting chemical warfare by reducing aggression in the males around them and increasing aggression in the females, which benefits them. This is Sobel's interpretation of the findings.

Huberman asks about the relationship between odors and hormones, particularly crying. Sobel says emotional tears are emitted in situations where nonverbal communication is critical. He notes that Darwin's The Expression of the Emotions in Man and Animals is influential, and that its chapter on tears is devoted to them. Darwin systematically explained the functions of emotional expressions but could not find a functional antecedent for tears, so he called them an epiphenomenon.

Sobel says the lab wondered whether the function could be a chemical signal. They harvested emotional tears and exposed participants to them. The tears are completely odorless, and people cannot detect them at all.

Huberman asks about generating smells in silico, using computers. Sobel says Google has tried to digitize smell, including an April Fools spoof, and other companies are attempting this as well. Sobel says his lab recently had what he described as a mini breakthrough, a paper published in Nature led by Aaron Raviv with major contributions from Kobe Snitz.

In the paper, the researchers developed an algorithmic framework that predicts the perceptual similarity of any two molecular mixtures with very high accuracy. They can also design mixtures with known similarities, which Sobel says is the more notable result. He says the work produced mixtures that smell the same without sharing any molecules, and that one of them smells like Chanel No. 5 but contains none of its components.

Sobel says the framework provides the infrastructure for digitizing smell, because it predicts similarity. He describes a map of 4,000 molecules for which the perceived smell of each is known. Any mixture can be projected onto this map, and its pairwise distance from all the points on the map measured. If it falls on lemon, the mixture smells like lemon; if it falls on tomato, it smells like tomato.

Sobel says this solved the problem of predicting the odor of any molecular mixture. The lab's next step is to find a set of components, which they call odor primaries, that can be used to mix any odor a person can perceive. He says this work is in collaboration with Jonathan Williams and the Max Planck Institute in Munich.

Sobel says Jonathan Williams, an atmospheric chemist skilled at using GCMS tools that measure molecules, measured odorants in Germany and transmitted the information to the Israeli lab over IP. The lab fed this into its algorithmic framework and recreated the odors from a device that mixes odor primaries. They tested four odorants in this proof-of-concept.

Rose failed: the recreated odor was perceived mostly as bubblegum. Anise also failed, and most people said it was cherry, which is not far off but still a failure. Gasoline was slightly but significantly better than chance at recreating it. Violets were the only one that clearly worked, with 15 of 16 people saying violets. Sobel says the first odor ever transmitted over IP was violets and cautions that this is only an initial proof of concept. He compares the current stage to black-and-white television in 1956 and says that recreating even something vaguely like wine would be an amazing success.