What this covers

Sean Carroll speaks with Ann-Sophie Barwich, a neuroscientist and philosopher, about why smell deserves a central place in our understanding of how the brain works. Barwich challenges the dominance of vision in neuroscience and philosophy of mind, arguing that smell is not a primitive or unreliable sense but a highly accurate, cognitively sophisticated system that forces a fundamental rethinking of perception itself. The conversation moves from the mechanics of olfaction—how the nose detects and codes chemical molecules—through the brain structures that process smell, to philosophical implications for how we understand rationality, decision-making, and the nature of perception. The discussion ranges across neurobiology, perceptual psychology, and epistemology, grounded throughout in concrete examples from wine tasting, body-odor chemistry, and clinical cases.

The core claim centers on why olfaction overturns the dominant Brain as Map/Mirror of Reality model. Unlike the visual cortex, which creates an orderly spatial map of the visual field, the olfactory bulb produces what Barwich calls a "chaotic mosaic" that varies across individuals and develops through experience rather than following a fixed genetic blueprint. This alone forces a reconception of how brains represent the world. Barwich further argues that variation in smell perception—whether someone tastes cilantro as fresh or soapy, or prefers certain wines—traces to objective physiological and genetic causes rather than mere subjectivity. She explores how smell operates through combinatorial coding, where single molecules activate multiple receptors and single receptors respond to many molecules, and how training can measurably lower detection threshold and thicken cortical tissue. The conversation also touches on how smell shapes unconscious choices in partner selection and disease detection, arguing that Biological Rationality often works through chemical signals the conscious mind never registers, which has implications for understanding free will and Post-hoc Rationalization.

Sharpest takeaway

Barwich argues that smell, far from being a primitive subjective sense, is a highly accurate, context-adaptive cognitive system whose study overturns the vision-centric 'brain as map/mirror of reality' model and reframes perception as active acquisition rather than passive representation.

  • Smell is harder to fool than vision because perception is grounded in direct chemical contact plus context, not just stimulus pattern
  • The olfactory cortex shows no orderly spatial map, unlike the visual cortex, forcing a rethink of how the brain represents the world
  • Perceptual variation in smell is traceable to genetics and developmentally-induced, environment-adaptive receptor patterns, so it is objective variation, not mere subjectivity

The argument · threads8 threads · 31 claims
0.85

Smell is more accurate and harder to fool than vision because it provides direct chemical contact with contextual information.

5 pointscentrality 5/5
  • There is no consistent mapping between molecular structure and smell quality: structurally similar molecules can smell alike or completely different, and structurally unrelated molecules can share the same smell quality — for example musks that are cyclical compounds versus nitrogen-containing ones share no structural similarity yet belong to the same quality group.

    “you also have the opposite, that you've got molecules very, very similar in structure with completely different smells, and also molecules that have completely unrelated kinds of microstructures, but very, very similar smells”

  • Rather than the brain mapping or mirroring reality, the brain measures what chemical features are present and in what gradient, concentration, and context, while the mind performs a contextual, behavioral, body-and-environment-tailored statistical evaluation of where you are and what you need — making smell more accurate than people assume and arguing against the picturing/mirroring metaphor of perception.

    “our brain measures basically what is there and in what kind of gradient, in what kind of concentration, in what kind of context... And the mind is much more in terms of... you basically have a contextual evaluation, you've got a behavioural measure that is tailored to your body, that is tailored to the environmental context, you've got a statistical evaluation of where you are and what you need”

  • The nose is harder to fool than the eyes and is more accurate, because while visual illusions exploit the fact that different retinal images can elicit the same mental image, smell puts you in direct contact with chemicals and incorporates context, making it an incredibly accurate measure of context rather than a fixed quality.

    “I think that the nose is actually harder to fool than your eyes. I think it's much more accurate, actually. It's very easy to fool your visual system through illusions”

  • Because of smell's nonlinearity, changing a single atom in a molecule can produce a completely different quality (e.g. pear to banana), and mirror-image enantiomers identical in shape can smell entirely different — the R and S carvones smell of caraway and mint respectively — unlike vision where a small change in photon wavelength looks almost identical.

    “with some molecule, if you change a small thing, it goes, for instance, from pear to banana, so a completely different quality”

  • Most of what people call taste is actually smell: the tongue's taste buds only detect salty, bitter, umami, sweet, sour (and now fat), while flavors like mint, strawberry, and vanilla are perceived through the nose when odor molecules released by chewing travel up the back of the mouth and throat to the olfactory epithelium, which is why a cold abolishes both smell and taste.

    “everything you call taste, when you taste something, or strawberry or mint or vanilla, etcetera, that is actually smell. And we know that because if you have a cold, you can't smell it. You can't smell and you can't taste a thing.”

0.79

Olfactory perception varies across individuals due to genetic diversity and neural plasticity, not subjectivity or illusion.

4 pointscentrality 5/5
  • The spatial pattern in the olfactory bulb is developmentally induced rather than genetically predetermined, so genetically diverse individuals have different neural representations of the same stimulus; this matters because the bulb is not encoding a fixed chemical map but a measure of contrast — what occurs more or less in the environment — allowing flexibility.

    “There is a huge difference. So you must imagine that when you have the bulb, you've got a spatial pattern that represents the chemical input. However, it's different in genetically diverse organisms.”

  • Perceptual variation in smell links to the extreme genetic diversity of the olfactory system, the genetically most diverse sensory system; for instance people who perceive cilantro as soapy or pungent rather than fresh carry a genetic mutation near an olfactory receptor gene, demonstrating that differing perceptions trace to material causes rather than mere subjectivity.

    “the olfactory system is one of the genetically most diverse systems... These are mutants in the sense like in... Like, kind of X-Men mutants, they have a genetic mutation near one of the olfactory's receptor genes.”

  • Smell's variation between and within individuals is not subjectivity (i.e. mere in-the-head perception with no grounding in reality) but objective variation that can be traced to and explained by fundamental physiological and genetic causes, dissolving the relativist worry that differing perceptions preclude objective truth.

    “the olfactory system is one of the genetically most diverse systems... These are mutants in the sense like in... Like, kind of X-Men mutants, they have a genetic mutation near one of the olfactory's receptor genes.”

  • Because humans inhabit and move between hugely diverse chemical environments, the nose changes the expression of its receptor patterns over time to accommodate different chemical surroundings, making smell context-adaptive in a way vision is not — vision stays roughly the same wherever you go while olfactory input depends on and reshapes itself to the environment.

    “after a while your nose actually changes the expression of receptor patterns to adapt to actually accommodate these different chemical environments. So you need a much more flexible system”

0.79

The olfactory system encodes context and behavioral meaning through combinatorial coding, not fixed chemical maps or quality hierarchies.

4 pointscentrality 5/5
  • Rachel Hertz's 2001 experiment gave people the identical butyric-acid mixture labeled either 'parmesan' or 'vomit' and they were adamant the smells differed and preferred parmesan; rather than proving smell is unreliable illusion, this shows both labels are correct because butyric acid occurs in both, demonstrating that a single promiscuous stimulus carries different behavioral meanings depending on context, which the olfactory system must resolve.

    “it was the same mixture, it was a mixture of butyric acid... it's often used as, "Oh, I see, this is why smell is so unreliable and it's just illusionary." No, not at all, because both parmesan and vomit do contain butyric acid.”

  • Olfaction uses combinatorial coding in which one molecule can be detected by many of the ~400 human receptor types and one receptor can detect parts of many molecules, so a single molecule may carry multiple qualities and most smells are mixtures of hundreds of molecules; this is why we can smell novel synthetic molecules that never existed in nature without evolving new receptors.

    “you've got one molecule can be detected by a variety of different receptors and one receptor can detect different parts of different molecules. So this is why we can smell so many different things”

  • Unlike the visual cortex, which maps the visual field in a one-to-one spatial correspondence (demonstrated by monkey studies where the seen image could be traced in the cortex), the olfactory system disperses its signal into a chaotic, disordered mosaic in the piriform cortex that differs across individuals, undermining the idea that the brain mirrors reality as an ordered map.

    “smell doesn't work like that. It's this mosaic. It just fans out, it's beautiful, but there is no order. And also, it's different across different individuals.”

  • Unlike vision where a point is one color, a single odor molecule can carry several qualities at once — for example cis-3-Hexenal smells of fresh cut green grass but also has leafy and earthy notes — and most everyday smells are mixtures of hundreds of molecules, with coffee containing about 800.

    “if I give you cis-3-Hexenal, which is one of my favorite molecules, it smells of fresh cut, green grass. But more than fresh cut green grass, it also has a leafy note, an earthy note... And that's just a single molecule. But most of the things you smell are mixtures, hundreds of molecules. Coffee has about 800 molecules.”

0.65

Perception is learned acquisition and augmentation of consciousness through training and attention, not passive representation of reality.

4 pointscentrality 4/5
  • Most of our understanding of mind and brain is based on the visual system because vision is immediate to consciousness and because we describe mental life in visual terms (images, pictures), but these concepts don't transfer to smell, which reveals we should rethink central concepts of perception and learn new things even about vision.

    “most of our understanding of mind and brain is based on the visual system, and for two reasons. One reason is that, well, we see things that's immediate to our consciousness... And also the way we talk about mental stuff, it's always about pictures, it's always about images”

  • Smell is a skill anyone with a normal nose can improve, not a freakish power: studies of people doing 20 minutes of smell training per day showed recordable increases in cortical thickness after just six weeks, and trainees both physiologically lowered their detection thresholds (smelling lower concentrations) and improved at identifying smells.

    “Doing smell training every day for 20 minutes and after six weeks, which is very, very fast, they could already record changes in the thickness of the cortex, so it really changes your brain.”

  • Focusing attention on and training the nose to recognize specific smells physiologically lowers the detection threshold for those familiar smells, so you literally learn to detect more of the world (molecules at lower concentrations) and cognitively obtain a more subtle picture — analogous to how training what to attend to in vision counters change blindness.

    “it does lower the threshold of detection. So it's smells you're familiar with and you train yourself to recognize, you have a lower threshold just physiologically to detect them, so you literally learn how to perceive more of the world”

  • An olfactory-centric view replaces the mirror metaphor of perception: perception is not a representation or mirroring of reality but an acquisition and augmentation of consciousness through constant evaluation and learning; because smells pop in and out of consciousness and must be re-identified across instances, smell makes vivid that what we take as our world is learned and acquired (not relative in an anything-goes sense, but not a passive picture either).

    “perception is not a mirror, it's not a representation of something, but it's an acquisition, it's an augmentation of consciousness, and it's a constant evaluation.”

0.65

Olfactory science poses fundamentally philosophical questions requiring integrated neurophilosophy and challenging visual-system-centered theories of mind.

4 pointscentrality 4/5
  • Hermann von Helmholtz argued in the 19th century that many visual images produce the same pattern on the retina and in the brain, so we have to move to figure out the differences between images, anticipating the modern understanding of why vision is fool-able.

    “in the 19th century, one of my favorite historical scientists, Hermann von Helmholtz, he actually made that claim already. He said, "Well, actually a lot of visual images have the same pattern on the retina and in the brain”

  • Olfaction reaches the cortex via only two synapses, the most direct sensory pathway, which led Ramón y Cajal to call it potentially the best model of the brain — by comparison two synapses don't even get you out of the retina in vision — yet decades of research show the apparent simplicity is deceptive and much remains unfigured out.

    “olfaction might be the best model of the brain you can have because it only has two synapses straight into the cortex... the most direct pathway for a sensory perception... two synapses don't even get you out of the retina and vision.”

  • While sympathetic to embodied cognition's emphasis on body-environment interaction, Barwich is critical of strong versions (e.g. Alva Noë's 'we perceive outside the brain'), holding a neuro-centric position aligned with Patricia Churchland: the body matters and must be understood to understand the brain, but the brain remains the central processing organ with causal primacy, as clinical cases (Oliver Sacks) of things going wrong demonstrate.

    “My issue with embodiment ideas is that they kind of take this brain almost also separate from the body... your brain is part of the body... but I don't think that means... that there's any form of causal primacy of a process.”

  • Philosophy of science should be revived as a 21st-century natural philosophy that is a productive producer of good experimental questions rather than merely an analyst, observer, and clarifier of models; following Patricia Churchland's radical neurophilosophy, contemporary scientific advances should generate new philosophical questions, and current olfactory science questions ('what are odors? what do they represent?') are themselves deeply philosophical.

    “to bring natural philosophy back into the 21st century and have a much more integrated role of philosophy, not just as an analysis and observer, but really actually as a producer of good questions.”

0.59

Smell unconsciously but rationally shapes major life decisions including partner choice and health judgments through biological information.

3 pointscentrality 4/5
  • Smell unconsciously shapes major decisions like partner choice: a body-odor study found women preferred the smell of men with complementary (genetically different) immune systems, but women on hormonal contraception reversed this to prefer men with more similar immune systems — a switch that may help explain divorce rates when couples stop contraception after marriage and a partner loses 'sex appeal.'

    “they preferred the smell of men who had a complementary immune system, so there was a slight genetic difference. And now comes the real shocker, because women who are on the pill, who had a hormonal change, changed their preferences, they had a preference for men or the body odor of men with a more similar immune system to themselves.”

  • Body odor carries information about another person's health, sickness, and dietary habits via metabolic skin emanations, and the human nose is susceptible to it though often unconsciously; this enables medical applications such as electronic noses that monitor body odor to flag changes warranting a doctor's visit, and cases like nurse Joy Milne who can correctly sniff out Parkinson's disease from patients' t-shirts.

    “It tells us also about the dietary habits of somebody, because the metabolic changes and the emanation from your skin, they tell you something about the health, the chemical emanations of another person.”

  • Following your nose is not irrational but a form of biological rationality, because smell reports hidden, non-visible features — whether something is rotten despite looking fine, like sour milk you smell rather than look at — so decisions influenced by smell can be rational utility-maximizing choices even when their true drivers are not consciously accessible, suggesting our self-attributed cognitive reasons often post-hoc rationalize choices made on factors we're unaware of.

    “there is a form of biological rationality and rationality in scare quotes to actually follow your nose in that sense, because it does tell you about the hidden things. The things that are not visible.”

0.44

Science advances by asking better questions and philosophy has been historically dismissed despite its value in experimental design and question formation.

5 pointscentrality 2/5
  • There is no intrinsic barrier between philosophy and science as practices; the divide is institutional, created by housing them in separate departments and even separate schools (humanities vs sciences), and by disciplines cultivating distinct languages and gatekeeping, which makes truly interdisciplinary work hard to fund and publish despite 'interdisciplinarity' being a buzzword.

    “I don't think there's an intrinsic barrier when it comes to philosophy as a practice and science as a practice... it's mainly the way we organize academia, we organize publications, it's very hard to get anything that is truly interdisciplinary-funded and published.”

  • Science is not the accumulation of facts but the practice of asking better questions; as knowledge grows like a sphere, the circumference of ignorance grows with it, so acquiring new skills and knowledge reveals ever more sophisticated things you don't know — a view embodied in Stuart Firestein's books on ignorance and failure and one where philosophical training in question-asking is an advantage for doing and designing experiments.

    “science is not the accumulation of facts, but the way to ask better questions. And the way we do that is really we're accumulating more and more knowledge, but with that knowledge, it's more like a sphere, you've got the circumference of ignorance, your knowledge grows... Your ignorance grows”

  • Genuine interdisciplinarity is not switching between disciplinary 'hats' but developing an integrated expertise that generates new questions through the integration itself; Dobzhansky's Genetics and the Origin of Species contained no new data yet became foundational to the evolutionary synthesis precisely because it integrated genetics and natural history into a new perspective that prompted new questions.

    “it's much more to develop an integrated expertise that goes beyond just having two different hats, and more in terms of what kind of future questions can we derive through the integration.”

  • Immanuel Kant called smell the most dispensable and ungrateful sense because it offers little pleasure given that there are more stinky things than good ones, exemplifying the long philosophical tradition of treating smell as the unimportant 'Cinderella of the senses.'

    “My favorite quote, actually, is by Immanuel Kant, who said, "Well, it's the most dispensable and ungrateful sense, and it doesn't give you much pleasure because there are many more stinky things than good ones."”

  • Recording a smell and sending it digitally is effectively impossible because reproducing it would require releasing exactly the same molecules in the same composition (e.g. coffee's ~800 molecules) on the receiving end, demanding a practically indefinite library of odorants; the only feasible 'Smell-O-Vision' is a device with a limited cartridge of predefined scents, which Barwich finds uninteresting.

    “You would have to release exactly the same molecules in that composition on the other hand, now imagine how many molecules are around, you would have to have an indefinite library of different odorants... So it's impractical, it's impossible”

0.34

Expert smell discrimination is a learned, analytical skill built through practice, not an innate talent or subjective preference.

1 pointcentrality 3/5
  • Expert wine tasting is not con-artistry or freakish ability but a highly cognitive, analytic, Sherlock-Holmes-style deductive process built over decades: the taster reads the wine's color and viscosity at an angle, swirls to release lighter notes, forms hypotheses against a learned mental library of wines, and falsifies candidates by checking tannin and acidity, while only naming about three descriptors because there is a biological cap on how many components we can identify instantaneously in a mixture.

    “this is an incredibly cognitive process, this is highly analytic and really kind of a Sherlock Holmes deductive way of perceiving.”