What the Nose Knows Before We Notice: Noam Sobel on Human Chemical Signals

1. Guest Background

Dr. Noam Sobel studies olfaction and chemical perception, including the neural mechanisms of human smell and the possibility that ordinary social behaviors help people sample one another’s odors. In this Huberman Lab Essentials conversation, Andrew Huberman asks him to connect those mechanisms with behavior, hormones, and the emerging ambition to digitize smell. The episode runs approximately 35 minutes and presents a concentrated tour of research questions rather than a complete account of any single experiment.

Sobel’s relevant experience comes into focus through the work he describes. He characterizes his laboratory as a human research group within a field that also studies other animals, and credits colleagues and students for projects on handshaking, the body odors of friends, and experimentally measured aggression. His perspective moves between the sensory apparatus, behaviors people may barely notice, and instruments designed to make those behaviors measurable.

That combination matters to the interview’s interpretation. Sobel can explain why a hand approaching the nose might be scientifically interesting, then discuss the controls needed to establish that sniffing occurred. The supplied background supports this research role and experience; it does not supply an institutional biography or a full career history. The useful introduction to this guest is therefore the range of questions his team investigates and the distinctions he draws between observed effects and explanations still under development.

2. What the Episode Covers

Sobel begins with the route from airborne molecules to neural activity. In orthonasal olfaction, inhaled molecules reach the olfactory epithelium and interact with receptors. Transduction converts that interaction into neural signals, which travel through the olfactory nerve to the olfactory bulb. His explanation places sniffing at the entrance to the process: the nose participates in sampling the material that the system will analyze.

Within the bulb, inputs from receptors of the same subtype converge at glomeruli. Sobel corrects the simplified description of a single convergence location, explaining that two glomeruli are involved. He illustrates the textbook model with hypothetical receptor types activated by coffee: their corresponding glomeruli would produce a pattern reflecting receptor identity. Yet this organizational principle does not finish the explanation of smell. Information travels onward to other brain regions, and Sobel explicitly describes coding beyond the bulb as murky. An account of the sensory input is not a complete account of recognizing an odor.

The handshaking experiment gives this discussion an everyday setting. The researchers noticed movements toward the face after handshakes, but face touching was already common enough to require a baseline. Participants first waited alone for three minutes while being recorded without knowing the recording had begun. A previously unfamiliar experimenter then entered and delivered a consistent explanation; half of these encounters included a handshake. Sobel reports an increase in movements bringing the hand toward the nose after handshaking.

The crucial additional measurement was nasal airflow. According to Sobel, participants sniffed as their hands approached their noses, strengthening the interpretation that these movements included olfactory sampling. They also smelled the hand that had not performed the handshake. He proposes comparison between self and other as a possible explanation, while the measurement itself establishes a narrower point: sniffing accompanied the movement.

The friendship work asks whether such chemical information might relate to immediate social affinity. Researchers recruited same-sex, non-romantic friends whose relationships had begun with a sense of clicking. Body odors were collected using a protocol involving unscented washing products and shirts worn during sleep. Sobel says both an electronic nose and human smell judgments found greater similarity among these friends than among random pairings. To investigate whether shared lives could explain that similarity, the team subsequently measured strangers before a nonverbal, round-robin mirror game and collected their evaluations of one another afterward.

3. Core Views: Reasoning, Examples, and Limits

The interview’s central interpretive challenge is to keep different outcomes separate. Sampling another person’s chemical traces, consciously recognizing an odor, liking that person, and forming a friendship are distinct events. The strangers’ mirror-game study makes this especially clear. Sobel reports that odor similarity predicted willingness to become friends and favorable social judgments, but did not predict how well participants coordinated their movements. The unsuccessful prediction limits the successful one. The findings support a relationship with particular immediate evaluations, while the interview does not describe evidence that these strangers later formed lasting friendships. Sobel speaks strongly about a causal contribution to friendship; the reported endpoints warrant a more specific conclusion.

His terminology offers another useful discipline. Sobel says his publications use “chemo signals” rather than “pheromone,” keeping attention on whether substances emitted by one person influence another. The practical scientific questions concern the source, exposure, receiver, and measured response. In his account of fear-odor research, several laboratories collected odors from the same people in fearful and non-fearful states and found that receivers could distinguish them or showed autonomic and behavioral effects. Describing fear as contagious conveys the idea, but does not establish that ordinary nervousness can always be detected or that smell provides a reliable reading of someone’s mental state.

The hexadecanal experiments complicate any attempt to label an odor with one behavioral meaning. Sobel traces the motivation to work on social buffering in mice and a receptor reported to be conserved across mammals. Those observations supplied a reason to investigate humans; they did not guarantee an identical human response. In aggression tasks comparing hexadecanal exposure with a control condition, he reports reduced aggression measures in men and increased measures in women. He says a further experiment conducted with functional brain imaging reproduced those directions. Within the account he gives, the same compound cannot sensibly be described as universally calming.

Sobel then offers an evolutionary interpretation. He describes hexadecanal as the most abundant semi-volatile substance from babies’ heads and suggests that decreasing male aggression while increasing protective female aggression could benefit offspring. This is a proposed functional bridge, not the outcome directly measured in the adult task. The account does not establish how naturally emitted infant quantities affect parents at home, or whether changes in a laboratory aggression measure translate into protective caregiving. His vivid image of babies conducting chemical warfare should retain its status as an explanation to investigate.

Emotional tears expose a different gap: conscious odor perception can diverge from a reported physiological response. After correcting himself, Sobel specifies men smelling women’s emotional tears. He says the participants perceived no odor, yet free testosterone fell by about 14 percent within roughly 20 to 30 minutes. Those figures are his account of an experiment, not a universal response to tears. He also reports an independent South Korean replication of the hormonal effect. His brain-imaging discussion includes reduced activity in regions such as the hypothalamus and fusiform gyrus, alongside an explicit admission that some of the changes remain unexplained.

The subsequent mouse example broadens the hypothesis without completing the human argument. Sobel describes research in which mouse-pup tears reduced adult male mouse aggression in a smell-dependent manner, then proposes a protective chemical blanket interpretation. The endpoints and species must remain visible. A change in human free testosterone and a reduction in mouse aggression are different observations. Together they motivate questions about function; they do not demonstrate that human tears prevent real-world violence.

The digitization discussion shifts from social interpretation to sensory engineering. Sobel says his team developed a framework for predicting the perceptual similarity of molecular mixtures and created olfactory metamers: mixtures without shared molecules that nevertheless smell the same. His example includes an alternative mixture resembling Chanel No. 5. The important principle is that reproducing an experience may not require reproducing the original molecular recipe. He describes a perceptual map of about 4,000 molecules and continuing work toward odor primaries that could be combined to reconstruct smells. Demonstrating perceptual substitutes and establishing a generally useful reconstruction system remain separate achievements.

The transmission test makes that distinction concrete. Measurements taken in Germany were sent over the internet and used to generate mixtures elsewhere. Sobel reports that the attempted rose smelled mostly like bubble gum, anise mostly like cherry, gasoline was recognized only slightly above chance, and violets were identified by 15 of 16 people. Violets provide an encouraging success, while the other targets expose uneven performance. The demonstration supports further development, with target-specific limitations that a single impressive recognition result cannot erase.

4. Learning and Application

A concrete way to use this episode is to read smell research through a sequence of questions: what exposure occurred, what the participant perceived, what the instrument measured, and what behavior or physiology changed. The handshake study provides a model. Establish the ordinary rate of face touching, compare encounters with and without a handshake, and add nasal airflow to determine whether the movement includes sniffing. Each step addresses an alternative explanation. The resulting evidence can support chemical sampling while leaving the proposed self-versus-other comparison open.

For social predictions, preserve both the outcome that worked and the one that failed. The strangers’ study is relevant to immediate favorable evaluations because those were measured. It supplies a different level of evidence for movement coordination, where the prediction did not succeed, and does not establish subsequent friendship quality. Someone considering an electronic-nose matchmaking application would therefore need evidence for that intended decision. Predicting a rating after one interaction is insufficient validation for selecting a long-term friend or collaborator.

The hormone and aggression examples suggest an equally practical rule: retain the population and endpoint whenever repeating a result. The tears account concerns men exposed to women’s emotional tears and a reported change in free testosterone. The hexadecanal account contains opposite directions for male and female aggression measures. Neither description supplies a usable intervention for managing a partner, child, or one’s own behavior. The interview does not establish the exposure conditions, doses, and real-world behavioral outcomes needed for such an application. Its immediate value is in sharpening questions about human chemical communication.

Missing methodological details should remain missing rather than being silently reconstructed. Sobel is uncertain about the duration of unscented-product use before the friendship samples, so the conversation should not become a precise home collection protocol. Similarly, the tears passage contains an effect estimate and a replication account without sample sizes or complete methods. A reader can identify what requires further checking, but cannot calculate general reliability from the reported percentage alone. Careful attribution preserves the finding’s interest without treating the interview as a full experimental report.

For developers, the digitization work encourages a clear choice of target. If the goal is recognizable sensory similarity, perceptual substitutes may be useful. If the task requires molecular fidelity, a metamer does not by itself meet that requirement. Evaluation should distinguish generating a detectable odor, approximating a category, and achieving correct identification. Reporting all four transmission targets helps expose the tradeoff between an encouraging demonstration and broad performance. Sobel himself compares the current ambition to early black-and-white television and explicitly warns against expecting a consumer device soon.

Medical applications require their own evidence. Sobel connects disease-related metabolic processes and metabolites with the possibility of future odor-based diagnosis, imagining an electronic nose operating continuously in a bathroom. He repeatedly places that scenario far in the future. When assessing such a proposal, the useful questions concern what the device actually detects and how that detection supports the claimed diagnosis. Reconstructing a recognizable violet odor does not establish disease detection. The episode offers a research direction and a framework for scrutinizing claims; the diagnostic capability would have to be demonstrated separately.

Source

• Original episode: Essentials: How Smells Influence Our Hormones, Health & Behavior | Dr. Noam Sobel