15 Chapter 4.8 — Olfaction: The Sense That Breaks the Plan
Chemoreception, the oldest sense, and the close of the unit
15.1 A sense that refuses the rules
We have spent an entire unit building up a plan. Receptors transduce one physical variable; the signal passes through first- and second-order neurons; it crosses the midline; it stops at a dedicated thalamic nucleus; it arrives in a primary cortical area where the world is laid out as an orderly map — retinotopy, somatotopy, tonotopy — and from there it feeds higher areas that pull out ever more abstract features. Vision obeys this plan. Touch obeys it. Hearing obeys it. Even taste, as we just saw, broadly obeys it, for all its kinship with the interior.
Olfaction obeys almost none of it.
Smell is the one major sense that does not route through the thalamus on its way to the cortex. It has no clean topographic map of the kind that organizes every other sensory cortex we have studied — and whether odors are spatially organized at all above the first relay is one of the genuine open questions of the field. It wires itself, directly and without the usual cortical intermediary, into the structures of emotion and memory. And it is, by a wide margin, the oldest of the senses — older than vision, older than hearing, older than the vertebrate body plan, older than neurons themselves.
In the unit overview I promised that when we reached the chemical senses, their differences from the shared plan would be “a feature to explain, not an inconvenience to apologize for.” This is where I make good on that promise. I want you to read olfaction’s rule-breaking not as a list of awkward exceptions but as a coherent portrait of a sense built on a different and more ancient logic than the others — a logic that makes sense the moment you remember where it came from. And I want to use it to close the unit, because olfaction returns us, at the end, to exactly where we began: a single cell in the sea, following a chemical gradient toward what it needs. The whole unit is a loop, and smell is how it closes.
15.2 The oldest sense
Begin with the deep history, because it explains nearly everything that follows.
At the very start of this unit, before any nervous system at all, we watched a single-celled alga steer toward light and an amoeba crawl up a chemical gradient toward food. That second trick — chemotaxis, sensing the molecular composition of the surrounding fluid and moving accordingly — is the ancestral sense, the one from which all the others are, in a sense, latecomers. A bacterium detecting a sugar gradient and swimming up it is performing the same fundamental operation that your nose performs when it catches dinner cooking: reading molecules in the medium and converting that reading into a behavior. Chemoreception is not one sense among several. It is the original sense, the thing sensing was before light and sound were ever pressed into service, and every nose, antenna, and taste bud on Earth is an elaboration of it.
This is why I find it fitting to close the unit here rather than open it. We began with single cells sensing chemicals because that is where sensing began; we end with the human version of that same primordial capacity, now grown into an organ that can distinguish a vast range of airborne molecules and route them into the machinery of memory and feeling. The chemical senses are not an afterthought tacked onto the end of a unit about the “real” senses of vision and hearing. They are the origin of the whole enterprise, met last only because we climbed up from the body’s interior and out to the distant world before circling back to the most ancient thing of all.
And the antiquity shows in the wiring. Olfaction’s strange architecture — its bypass of the thalamus, its direct hooks into the limbic system — is not a flaw or a quirk. It is old plumbing. The olfactory system was routing chemical information into the control of behavior long before the thalamus evolved into the grand gateway-to-cortex it became for the younger senses. Smell never adopted the newer arrangement because it never needed to; it had already solved its problem, in its own way, before the standard plan existed. When you understand olfaction as the ancient sense still running on ancient infrastructure, its violations of the plan stop looking like exceptions and start looking like seniority.
15.3 A distance sense built on a different physics
Before we go inward to the wiring, one point about what kind of sense smell is, because it matters for the control-theory spine of this unit.
Olfaction, unlike taste, is a distance sense. Its partner senses the chemical already in the mouth, at the threshold of the body, with no lead time at all — taste is reactive, a gatekeeper. But smell reaches out, catching molecules that have travelled to you through the air from a source that may be far away and not yet doing anything to you: the predator upwind, the smoke of a fire over the hill, the rot you should not step in, the meal three rooms away. In the language we built across this unit, smell buys lead time. It belongs with vision and hearing on the predictive, allostatic end of the gradient — a sense that lets you detect a situation before it becomes an emergency, and therefore lets you act in advance.
But — and this is the part I want you to appreciate — it buys that lead time on a completely different physical principle from the other distance senses, and that difference has consequences. Vision and hearing ride on waves: light and sound propagate in clean, fast wavefronts that travel in straight lines and arrive in a definite direction at a predictable speed, which is exactly why those senses can localize a source so precisely and so quickly. Smell has no such luxury. Odor molecules travel by diffusion and, mostly, by the turbulent churning of air — they arrive in ragged, intermittent plumes, swirled and broken up by every gust and eddy, concentrated here and absent there a moment later, with no reliable relationship between where the molecule arrives and where it came from. The result is a distance sense with a strange and difficult temporal structure: smell tells you, often quite reliably, that something is out there, but it is far worse than vision or hearing at telling you precisely where, and it delivers its information in fits and gusts rather than a steady stream. An animal tracking an odor to its source does not aim at it the way you aim your eyes; it casts about, sampling the broken plume, edging up the gradient — much closer to the amoeba’s patient chemotaxis than to the eagle’s snap of attention. So olfaction is a predictive distance sense, yes, and it earns its place on the allostatic end of the spine — but it is a distance sense that never got the clean spatial precision of the wave senses, because the physics of drifting molecules will not allow it. The deep wiring to memory and value, which we are about to meet, is in part how olfaction makes up for this: if you cannot pin down where the smell is, you had better be very good indeed at knowing what it means.
15.4 How smell works: the receptors and the combinatorial code
Now the mechanism, which is genuinely beautiful and which earned a Nobel Prize, and which — this is the satisfying part — has at its heart the same molecular family we met in the taste chapter, deployed on a far grander scale.
The receptors for smell are G-protein-coupled receptors, the same broad class that detects sweet and bitter on the tongue and catches photons in the eye. But where taste has a small handful of these and vision has a mere four (one for dim light, three for color), olfaction has hundreds. The discovery of this enormous receptor family by Linda Buck and Richard Axel in 1991 — for which they shared the 2004 Nobel Prize — opened up the molecular understanding of smell, and one of their findings was startling enough to reshape how people thought about the sense: the genes encoding odorant receptors constitute the largest gene family in the mammalian genome [@buck1991novel; @nobel2004].
How large, exactly, is worth being careful about, because this is a place where a memorable number gets repeated more confidently than it should be. The human genome contains on the order of 800 to 900 olfactory-receptor genes — but roughly half of these are pseudogenes, broken relics no longer producing a working receptor, leaving something like 400 functional odorant-receptor genes in a typical person [@niimura2003human; @malnic2004human; @olender2020annotation]. You will often see the figure “about a thousand genes, about 3% of the genome,” and it is a useful order-of-magnitude anchor, but notice that it quietly conflates the total count (including the dead pseudogenes) with the functional count, and that the precise numbers depend on how one annotates a borderline gene. The honest statement is a range with the right caveats: hundreds of functional receptors, drawn from a gene family of close to a thousand members, the largest in the genome — and even after the heavy losses, far more receptor types than any other sense commands.
The fact that roughly half our olfactory-receptor genes are pseudogenes is not an accident of bookkeeping; it is an evolutionary signature, and it tells a story worth telling. Compare us to a mouse, which has a larger functional repertoire — well over a thousand working receptors — and a much smaller fraction of pseudogenes [@niimura2005evolutionary]. The human lineage has been losing functional olfactory-receptor genes, letting them decay into pseudogenes, faster than other primates [@gilad2003human]. The usual reading is that as our ancestors came to lean more heavily on vision — the primate bargain, trading the nose for the eye — the selective pressure maintaining the full olfactory repertoire relaxed, and genes no longer earning their keep were free to rot.
But I want to flag two cautions, in keeping with this book’s habits. First, the “humans are visual animals with feeble noses” story is easy to overstate; humans are in fact remarkably good at detecting and discriminating odors when actually tested, and the decline in gene number does not translate into the perceptual incompetence the folklore implies. Second — and this is one of the refrains of this book — resist the temptation to spin the gene loss into a triumphant just-so story about human specialness. We did not “rise above” the chemical world by shedding smell genes. A lineage simply stopped paying for hardware it was using less, the way the sea squirt stopped paying for its swimming apparatus once it settled down. Gene loss is loss, not ascent. It is a fact about relaxed constraint, and it deserves to be told as one.
Now the genuinely clever part, the insight that explains how a few hundred receptor types can represent an enormous range of distinct smells. Two rules do the work [@buck1991novel; @malnic1999combinatorial]:
First, each olfactory sensory neuron expresses just one type of receptor. Of your several hundred receptor genes, any given neuron in the nose picks one and commits to it — a remarkable instance of cellular discipline whose mechanism is still not fully understood. So the nose contains several hundred populations of neurons, each population defined by the single receptor it carries.
Second — and this is the key — the receptors are not narrowly specific. A given odorant receptor does not recognize one and only one odor molecule. Instead, it responds to a molecular feature — a particular shape, a particular chemical group — that many different molecules might share. And conversely, a single odor molecule typically possesses several such features and therefore activates several different receptors at once. The relationship between molecules and receptors is many-to-many: one receptor, many odorants; one odorant, many receptors.
Put those two rules together and you get a combinatorial code, and it is exactly the right design for the problem. Each odor is represented not by which single receptor fires — that would cap the system at a few hundred distinguishable odors, far too few — but by the combination of receptors it activates, the particular chord struck across the receptor array. Think of how 26 letters generate an unbounded vocabulary, or how three types of cone let you see a vast range of colors: a modest set of elements, combined, explodes into enormous representational capacity. With a few hundred receptors each either active or not, the number of distinct patterns the array can in principle represent is astronomically larger than the number of receptors. Now, how many of those representable patterns the system actually resolves into separable percepts is a different and much harder question — one we will see is genuinely unsettled in a moment — so I want to be careful to claim only what the combinatorial argument establishes: not that we do discriminate some specific huge number of odors, but that the coding scheme removes the obvious bottleneck, giving a few hundred detectors the representational room to stand for far more odors than there are detectors. The code is in the combination. What that buys in actual discriminable smells is a separate matter.
You will sometimes read that humans can distinguish about 10,000 odors. That figure is old, of uncertain origin, and was essentially a back-of-the-envelope guess that hardened into a textbook “fact” through sheer repetition — already a lesson worth noting. In 2014 a widely-publicized study went the other way and claimed the true number was at least one trillion distinguishable smells [@bushdid2014trillion]. That number swept the popular press — and then other researchers reexamined the underlying mathematics and argued that the method was unsound, that the same logic applied to other senses would yield absurd results, and that the data simply do not license any specific large number at all [@meister2015trillion; @gerkin2015trillion].
I tell you this not to settle on a figure — there isn’t a defensible one — but because the episode is a small masterclass in scientific self-correction and in the §-discipline this book keeps urging. A striking, quotable result (“a trillion smells!”) swept through the press; careful reanalysis showed the striking number was an artifact of the method; and the honest current position is we do not have a principled count. The right answer to “how many smells can we distinguish?” is “a great many, and we cannot yet put a trustworthy number on it” — which is less exciting than a trillion and considerably more honest. When a suspiciously round or suspiciously enormous number attaches itself to a hard measurement problem, treat it as a flag, not a fact.
15.5 Into the brain: convergence, a map in the bulb — and a different code in the cortex
Here is where olfaction’s architecture starts to diverge from everything else in the unit, and the divergence happens in two stages that are worth separating, because the contrast between them is the conceptual heart of the chapter.
Stage one: the nose builds a map. The axons of all those scattered sensory neurons travel back to the olfactory bulb, the first relay, sitting just above the nasal cavity beneath the front of the brain. And there, something orderly happens. Recall that the neurons are scattered across the nasal epithelium, each carrying one of several hundred receptors, all jumbled together. As their axons reach the bulb, they sort themselves: all the neurons expressing the same receptor converge onto the same one or two targets in the bulb — small spherical knots of neuropil called glomeruli [@mombaerts1996visualizing; @nobel2004]. Each glomerulus thus becomes the dedicated collection point for one receptor type, gathering the signal from that whole population of like-receptored neurons scattered across the nose. And the positions of these glomeruli are roughly reproducible from one individual to the next. So the bulb does contain a kind of map — an orderly, repeatable spatial array in which each receptor type has its place, and an odor, by activating its particular combination of receptors, lights up a particular constellation of glomeruli, a spatial signature you could in principle photograph. So far, this is recognizably plan-like: receptors sorted into an orderly cortical-style map, each odor a pattern across it.
Stage two: the cortex stops preserving the map. Now follow the signal one synapse further, from the bulb into the olfactory cortex proper — chiefly the piriform cortex, the main destination — and the orderly array of the bulb is not carried forward. The neat glomerular map does not project onward in a preserved, point-for-point fashion the way the retina’s map is preserved into V1 or the cochlea’s frequencies are preserved into A1. Instead, each region of piriform cortex receives a scattered, distributed sampling of inputs from all across the bulb, so that a given odor activates a sparse, distributed set of piriform neurons that — when one sorts them by odorant identity — show no obvious spatial clustering: neurons responding to a given smell are not grouped together but salted across the cortical sheet [@stettler2009representations]. I want to state this carefully, because the strong version is easy to overstate and the chapter will turn on the distinction in a moment. What is established is that the bulb’s identity-based map is not preserved into piriform — piriform codes odor identity in a distributed, non-topographic way. What is not established is the stronger claim that piriform has no spatial organization of any kind. Those are different statements, and keeping them apart is the whole game here. So: the map the bulb assembled — a map of which receptors an odor activates — is, at the next stage, abandoned. Whether some other organizing principle takes its place is a question we are about to take seriously.
I want to dwell on this because it is the cleanest possible illustration of why olfaction “breaks the plan,” and because the contrast with the sense we just studied is so sharp. Recall from the taste chapter that one of the live debates was whether taste cortex has a spatial map, and that the recent weight of evidence was against one — that taste, too, seems to use a distributed rather than a topographic cortical code. Olfaction is the same lesson written even larger and with less ambiguity: here we can watch a perfectly good map of receptor identity get built in the bulb and then not preserved one synapse later in the cortex, which codes that same identity in a distributed form instead. Vision, touch, and hearing preserve their maps all the way up because the things they represent — position, position, frequency — are intrinsically spatial or continuous dimensions for which a map is the natural format. But what would a map of smell even be a map of? Odor “quality” does not vary along one or two obvious continuous dimensions the way visual space or sound frequency does; the space of possible smells is bewilderingly high-dimensional, and molecules that smell alike need not be similar in any simple physical or chemical respect.
It is tempting to conclude from this that smell simply has nothing for a map to organize — that it is intrinsically too high-dimensional and unruly to be laid out on a cortical sheet. But I think that conclusion is too quick, and the more honest and more interesting possibility is that we may have been looking for the wrong map all along. When we test for a spatial organization of smell, we almost always sort by chemical identity — we ask whether “this odorant” and “that odorant” activate neighboring patches of cortex, and we find scatter, and we conclude “no map.” But a sensory sheet can be organized by a variable we never thought to plot against. We may keep failing to find the map not because there is none, but because we keep testing the wrong metric.
And there is a specific, well-motivated candidate for the metric that might matter, advanced most forcefully by Noam Sobel and colleagues: valence — the pleasantness or unpleasantness of a smell. The evidence that valence is the right axis to try is genuinely strong. When researchers take large datasets of how people describe and rate odors and ask, statistically, what single dimension captures the most variation, the answer that comes back — reproducibly, from independent groups working on independent data — is pleasantness. The first principal component of human odor perception is hedonic: how good or bad the smell is, outranking intensity and every chemical-structure descriptor [@khan2007pleasantness; @arzi2011compass]. Pleasantness is even partly predictable from a molecule’s physical structure, and it is the second axis, edibility, that tends to come next — which is to say that the dimensions along which we actually experience smells are dimensions of value to the body, not dimensions of chemistry. Sobel’s group put the proposal in exactly the terms this section needs: use perceptual pleasantness as a compass for finding the olfactory neural map — let the dominant axis of the percept tell you which axis to look for in the brain [@arzi2011compass].
Here I have to hold the line on calibration, because this is precisely the kind of elegant idea that the gustotopy story should have taught us to handle carefully. That valence is the leading perceptual dimension, and that hedonic information can be read out from activity in the olfactory and orbitofrontal cortices, is well-supported. That there is a literal spatial, topographic valence map laid across olfactory cortex is a further claim that has not been cleanly established — decodable from a distributed pattern is not the same as arranged in an orderly map, and odor valence is also far more learned and culture-dependent than taste valence, which sits awkwardly with any hard-wired hedonic chart. So the responsible statement is the asymmetric one: the compass points convincingly at valence, but no one has yet found, with confidence, the spatial map it is pointing toward. What I want you to take from this is not a settled map but a reframed question — and the reframing is the payoff. The dead-end version (“smell is too high-dimensional to map, full stop”) becomes a live and thesis-relevant one: perhaps the chemical senses are organized not by what the molecule is but by what it means for the body — and we have been missing the structure because we kept asking the first question when the brain may be answering the second. The map was perhaps never going to be a map of odor identity. It may be a map of value — and value, as the whole unit has insisted, is what sensing is ultimately for.
There is a sharper, more concrete version of this idea worth giving briefly, because it grounds the abstract word “valence” in real chemistry and connects it to a system we have already met. Look at what sits at the aversive end of odor space, and it is not a random scatter of unpleasant molecules — it is, to a striking degree, the chemical signature of decay. The smells we find most reliably and most universally revolting are the volatile products of bacterial decomposition: the diamines putrescine and cadaverine (named, respectively, for putrefaction and for the corpse), released as bacteria break down the amino acids of rotting flesh; the sulfur compounds of decomposing protein; the indole and skatole of feces. These are the molecular markers of food that has turned and matter that will sicken you, and our aversion to them is about as close to a human universal as olfaction offers. Tellingly, some of these decay molecules are detected not through the ordinary combinatorial odorant-receptor system at all, but through a separate, evolutionarily ancient receptor family — the trace amine-associated receptors — that across vertebrates is tuned to exactly these amines of rot and predator and danger, and that drives innate avoidance [@izquierdo2018diamine]. The aversive pole of smell, in other words, may have something close to its own dedicated channel — a labeled line for rottenness — rather than being merely one neighborhood of a general-purpose odor space.
This unifies the chemical senses around a single ancient job. The rotten-smell-to-avoidance arc is the distance-sense twin of the bitter-taste-to-rejection arc from the last chapter — both are the body asking is this safe to take in?, one at arm’s length and ahead of commitment, the other at the threshold of the mouth. And it is the natural entry point for disgust, the affective system we glimpsed converging on the insula in the pain chapter: in its evolutionary core, disgust is very plausibly a food-rejection program — the curled lip and wrinkled nose are a literal motor command to expel and to recoil — and the smell of rot is its most direct trigger.
And here the insula thread that has run through these chapters closes in a way that reaches forward to the social brain. In a now-classic experiment, participants both inhaled foul odorants and watched video of other people making the facial expression of disgust at a bad smell — and the same region of the anterior insula lit up in both conditions, for disgust specifically and not for pleasant odors or happy faces [@wicker2003disgust]. The authors’ point is exactly the safe-to-eat logic of this chapter, now running socially: in a world where spoiled food is a real threat, watching someone recoil from a dish lets you infer that food is bad — do not eat it without having to smell the rot yourself. Disgust read off another’s face and disgust felt in your own nose draw on overlapping cortex. Notice that this is the precise twin of a finding you have already met for pain — that witnessing pain in another and feeling it yourself engage many of the same regions — so disgust and pain turn out to be parallel all the way down: both are body-protecting valence signals, both converge on the insula, and both can be triggered by another’s state as well as one’s own. (How literally to read this as a “shared representation” or a mirroring mechanism is a genuine and contested interpretive question — overlapping activation is common territory, not proof of identical neurons — and we will take it up with the rest of the social-neuroscience material later in the course.) For now the point is only that the insula, which we have watched gather pain, taste, and flavor, gathers this too, and even gathers it secondhand.
We will take disgust up properly in Unit V and in the social-neuroscience lectures, where the machinery of valuation, emotion, and empathy belongs; here I only want you to see that the limbic shortcut, the smell-of-decay, and the insular valence thread are all the same story from different angles, the chemical senses doing the oldest job there is. But I will not overstate the axis, because its symmetry fails: the aversive end of odor space is chemically anchored, cross-culturally stable, and possibly hard-wired, but the pleasant end is a far looser and more learned affair — jasmine and woodsmoke and coffee are not pleasant because they are “edible” or “fresh,” and one culture’s delicacy is another’s stench. So the honest shape of the axis is lopsided: a sharp, ancient detector of rot at one end, and a soft, heterogeneous, culturally-elaborated miscellany of pleasures at the other. That asymmetry is itself a clue — it is exactly what you would expect if the deep, conserved function of olfactory valence were the avoidance of spoilage, with the appreciation of perfume a later and more optional overlay.
15.6 The limbic shortcut: smell, memory, and emotion
Now the other great peculiarity, and the one with the richest human resonance: where olfaction goes, and how directly it gets there.
Recall the standard plan once more: a sense reaches the cortex by way of the thalamus, that obligatory gateway. Olfaction is the great exception. The piriform cortex receives its input directly from the olfactory bulb, without an intervening thalamic relay — smell is the one sensory stream that reaches its primary cortex without first passing through the thalamic gate that all the others must clear. (There is a thalamic pathway in olfaction, projecting onward to the orbitofrontal cortex, and it matters for the more deliberate, attentive aspects of smell — so the textbook line “olfaction completely bypasses the thalamus” is a slight overstatement worth correcting. But the primary cortical pathway, bulb to piriform, genuinely does skip the thalamus, which no other sense does.)
And look at where the olfactory bulb’s targets sit. Beyond piriform cortex, the bulb projects directly to the amygdala — the structure we will study as central to emotion, and to fear above all — and to the entorhinal cortex, which is the principal gateway into the hippocampus, the structure most associated with the formation of new memories. The chemical senses, in other words, have a direct line into the brain’s emotional and mnemonic machinery that no other sense enjoys so immediately. Vision and hearing reach the amygdala and hippocampus too, of course, but by longer, more processed, more cortically-mediated routes. Smell has a shortcut.
This anatomy is the solid, defensible core of a famous and much-romanticized idea — that smell is peculiarly bound up with memory and emotion, that a single chance odor can summon a flood of feeling and a vivid scene from decades past. This is the “Proust phenomenon,” after the celebrated passage in which the taste-and-smell of a madeleine dipped in tea unlocks the narrator’s childhood in a rush of involuntary recollection. And here I have to do the thing this book always does with a beloved story: separate the part that is well-supported from the part that is overstated, because both are present and students deserve to know which is which.
The well-supported part: odor-evoked memories really do appear to be reliably more emotional and to carry a stronger feeling of being transported back to the original time and place than memories cued by words or pictures, and when people recall such memories the amygdala is more strongly engaged [@herz2004neuroimaging; @herz2016role]. That much has real behavioral and neuroimaging support, and it fits the anatomy of the limbic shortcut handsomely. The overstated part: the popular slogan that smell is simply “the memory sense,” or that odor memories are more accurate or more numerous than other memories, goes well beyond the evidence — odor-cued memories are not generally more accurate, and when smell is compared not against weak cues but against other emotionally potent cues such as music or faces, its supposed uniqueness gets considerably more nuanced and harder to pin down. So the honest statement is a middle one: olfaction genuinely does have a privileged anatomical line into emotion and memory, and odor-evoked recollections genuinely do tend to come with more feeling — but “smell is the memory sense, full stop” is folklore that runs past what has actually been shown. The shortcut is real; the mythology around it has outrun the data.
The Proust phenomenon is doing real load-bearing work in this section (it’s the human payoff of the limbic-shortcut anatomy), and it is genuinely shaky in its strong form, so per §5A I’ve written it in but split it explicitly into supported core vs. overstated folklore rather than either debunking or endorsing wholesale. My read of the literature: the anatomy (direct bulb → amygdala/entorhinal projection, thalamic bypass) is solid and exempt; the behavioral claim that odor memories are more emotional/vivid has decent support (Herz’s cross-modal work); but the studies are mostly modest-N, the effect is on subjective emotionality rather than accuracy, and the crucial comparisons against other privileged cues (music, faces) are thin and mixed. I’ve hedged accordingly. If you want this firmer or softer I can move it either way — but I’d resist stating the strong “smell is THE memory sense” version, which is where most textbooks and all of the popular literature land, and which I think fails the §4A bar. Flagging rather than cutting because the anatomy genuinely earns the discussion; it’s only the folk-amplified version that doesn’t.
I have deliberately kept human pheromones out of the main text, and want to flag the decision rather than make it silently, since it’s the kind of charming-but-shaky material the transcript-era lecture might have included. The defensible facts: the human vomeronasal organ (VNO) develops in utero and then regresses, adults have no functioning accessory olfactory bulb, and the V1R/V2R vomeronasal-receptor genes and the TRPC2 transduction channel are pseudogenized in humans — solid comparative anatomy/genetics, exempt under §4A, and I could add a sentence or two on it as a nice “another thing humans have lost” companion to the OR-pseudogene story if you want it. But the behavioral “human pheromone” literature (androstadienone/estratetraenol “signaling” masculinity/femininity, putative menstrual-synchrony effects, etc.) is exactly the §4A trap: effects are small, context-dependent, mood-level rather than physiological, inconsistently replicated, and currently being re-amplified by popular science writing (I saw fresh 2026 blog pieces claiming “the consensus is crumbling”). Direct tests — occluding the VNO and showing it makes no difference to the perception or neural processing of putative pheromones — argue the VNO has no function and that any effects run through the main olfactory system. My recommendation: either omit entirely (current choice) or include only the regressed-VNO/pseudogene anatomy as a brief aside, explicitly noting that claims of functional human pheromones remain unsubstantiated. I would not present human pheromone communication as established or even as a live even-money debate; the weight is clearly on the skeptical side. Your call on whether to add the anatomy aside.
15.7 Flavor: smell pretending to be taste
There is one everyday phenomenon that pulls together both chapters of the chemical senses, and it makes a point this unit has been building toward since the vision chapters, so it earns a place here even though part of it points forward to Unit V.
Ask yourself what you are tasting when you eat a peach. You will be tempted to say the peachiness is a taste — but it cannot be, because we established in the last chapter that the tongue has only a handful of channels: sweet, sour, salty, bitter, umami. There is no “peach” receptor on the tongue and could not be. The sweetness of the peach is taste; everything else about it — the entire rich, specific, unmistakable peach-ness — is smell. What we casually call the “taste” of food is mostly flavor, and flavor is a construction, a multisensory percept the brain assembles out of taste (from the tongue), aroma (from the nose), and even touch and temperature and the trigeminal bite of chili or the cool of mint.
The crucial and slightly magical fact is that the smell contributing to flavor does not arrive the way you think it does. When you chew and swallow, volatile molecules are driven up and backward, from the mouth into the nasal cavity through the passage at the back of the throat, reaching the same olfactory receptors from behind. This is retronasal olfaction — smelling through the back door — as opposed to the ordinary orthonasal smelling of sniffing the air through your nostrils [@rozin1982taste; @small2005differential]. And here is the part that reveals how thoroughly the brain constructs perception: the very same odor molecule produces two different conscious experiences depending on which way it arrives. Sniffed orthonasally, it is perceived as a smell, located out in the world — “this room smells of coffee.” Released retronasally from food in the mouth, the identical molecule is perceived as a component of taste, located in the mouth — “this coffee tastes rich.” The brain even refers the sensation to the wrong organ: it insists the experience is happening on your tongue when the receptors doing the work are up in your nose. And this referral is cognitively impenetrable — it persists even once you understand exactly what is going on, which is the signature of a genuine perceptual construction rather than a mere belief.
You can prove all of this on yourself in ten seconds, and you have probably done it by accident every time you have had a cold. Pinch your nostrils shut and eat something — a jellybean is the classic demonstration. With the retronasal airflow blocked, the flavor collapses: you get sweetness and texture and not much else, a sad gray ghost of the food. Release your nose mid-chew and the flavor floods back, instantly and vividly, and — this is the tell — it floods back into your mouth, not your nose. This is why food is so disappointing when you are congested. You have not lost your sense of taste at all; your tongue is working perfectly. You have lost the smell half of flavor, and with it most of what made the food worth eating. People who permanently lose their sense of smell — through head injury, or, as a great many people learned firsthand in recent years, through viral infection — frequently describe it as a loss of taste, and report that eating has become joyless, precisely because flavor was mostly smell all along and they never knew it.
And notice — this is why the section belongs in this chapter and not only later — that flavor gives us one more convergence on the cortex that has been the through-line of these two chapters. Retronasal odor, the smell that masquerades as taste, turns out to be processed differently in the brain from the very same odor sniffed orthonasally: retronasal food-odor recruits the insular gustatory cortex, the taste region, in a way that orthonasal odor does not, and silencing that taste cortex selectively disrupts the flavor role of smell while leaving ordinary sniffed smell intact [@blankenship2019retronasal]. So the insula — pain in one chapter, taste in the next, and now the binding-site where smell becomes flavor — collects yet another of the body’s evaluative chemical signals. Flavor is, in the end, the brain deciding how good this food is, and it builds that verdict in the same affective-chemical cortex where pain and taste already converged.
What I am deferring to Unit V is the part of flavor that is about reward rather than construction: why flavor drives us to eat, how the orbitofrontal cortex assigns it value, why the fifth forkful of a food is less pleasurable than the first (sensory-specific satiety), how all of this steers feeding. That is a valuation story, and valuation is Unit V’s business. Here I want only the perceptual point — that flavor is a multisensory construction, mostly smell, assembled by the brain and referred to the mouth — because it completes the chemical senses and because it pays off, in a chemical key, the constructive-perception lesson that the vision chapters taught in a visual one. The brain does not receive flavor. It builds it.
15.8 Closing the unit: from a cell in the sea to a verdict in the mouth
Let me end this chapter, and with it the unit, by standing back.
We began Unit IV with a question: what does an animal need to sense in order to act? And we began the answer with the simplest possible case — a single cell in ancient water, with a light-sensitive patch and a chemical-sensitive membrane, steering itself toward what it needed and away from what would harm it. Receptor, coupling, effector. Everything since has been elaboration of that primordial loop. We followed it outward and upward: into the body’s own interior (the interoception of Unit II, the prime mover that sets behavior going); out to the surface of the skin (somatosensation, the world in contact, reactive, no time to think); through the alarm of pain (the sense that something is bad); and then out to the great distance senses — vision and hearing — that reach across space and therefore across time, buying the lead time that lets a brain predict instead of merely react. The whole unit ran inside-out and reactive-to-predictive, from the chemistry of the blood to the light of a far hillside.
And now, at the end, the chemical senses have closed the circle. Taste returned us almost all the way to the inside — chemoreception at the very threshold of the body, the gatekeeper deciding swallow-or-spit, reporting through the visceral hub of the brainstem to the affective cortex of the insula, the most ancient control decision of all dressed in vertebrate anatomy. And olfaction returned us, at the last, precisely to where we started: a sense built on reading molecules drifting through a medium, the same operation the amoeba performed, now grown into an organ that reaches out across distance to tell an animal what is out there and — through its ancient shortcut into memory and feeling — what it means. We end the unit performing, with a few hundred receptors and a combinatorial code and a direct line into the limbic brain, the very first thing sensing ever did: following the molecules toward what we need.
There is a pleasing symmetry in that. The unit opened with a cell sensing chemicals and closes with the human elaboration of chemical sense; it opened by insisting that sensing exists for the sake of action and closes at the doorstep of the unit where action takes center stage. For across all of it — the touch and the light and the sound and now the smell and the taste — the senses were never cameras pointed at reality for its own sake. They were the front end of a control loop whose business is doing: move toward this, away from that, swallow this, spit that out. We have now built the sensing. In Unit V, the brain stops merely detecting that something is good or bad and starts acting on the verdict — selecting, valuing, moving, deciding. The insula, which we have watched gather the body’s chemical and painful verdicts across these last chapters, will be waiting there as one of the doorways from sensing into value. The world has been re-projected inward, in all its modalities. Now we find out what the brain does with it.
Reasonably settled:
- Chemoreception is the evolutionarily oldest sense, predating neurons and the vertebrate plan; olfaction’s unusual architecture reflects ancient infrastructure, not malfunction.
- Olfactory receptors are GPCRs and form the largest gene family in the mammalian genome (~800–900 genes in humans, ~half pseudogenes, leaving ~400 functional); humans have lost functional OR genes faster than other primates.
- Smell uses a combinatorial code: one neuron expresses one receptor; receptors are broadly tuned (one receptor–many odorants, one odorant–many receptors); odors are coded by the combination of receptors activated. Same-receptor axons converge on shared glomeruli in the olfactory bulb (Buck & Axel; 2004 Nobel).
- The olfactory bulb contains a reproducible glomerular map, but the primary olfactory (piriform) cortex represents odors as distributed, non-topographic population codes — the map is not preserved onward.
- Smell is the one sense whose primary cortical pathway (bulb → piriform) bypasses the thalamus, and the bulb projects directly to the amygdala and entorhinal/hippocampal memory circuitry — a limbic shortcut no other sense has.
- The anterior insula represents disgust whether a foul smell is inhaled or observed in another’s face (Wicker et al. 2003) — the disgust twin of the seeing/feeling-pain convergence, and the social extension of the “is this safe to eat?” logic. (The activation overlap is robust; the strong “shared representation”/mirror interpretation is contested.)
- Flavor is a multisensory construction (mostly retronasal smell, plus taste, touch, temperature); the same molecule is perceived as “smell” orthonasally and as part of “taste” retronasally, and the percept is referred to the mouth. Retronasal odor recruits insular taste cortex; orthonasal does not.
Genuinely unsettled, and presented as such:
- Whether odor quality has any spatial/topographic logic in the cortex above the glomerular array. The current picture is distributed/non-mapped piriform coding for odor identity. A leading reframe (Sobel and colleagues) is that we may be testing the wrong metric: valence (pleasantness) is reproducibly the first principal component of odor perception, making it the leading “compass” for any neural organization. But valence being the dominant perceptual axis — and decodable from hedonic-network activity — is well-supported, whereas a literal spatial valence map in olfactory cortex is not established. The axis is also asymmetric: the aversive pole is chemically anchored (decay products — diamines, sulfur compounds, indole/skatole — some detected via a dedicated TAAR channel that drives innate avoidance) and cross-culturally stable, while the pleasant pole is heterogeneous and largely learned. The deep conserved function of olfactory valence is plausibly spoilage/contamination avoidance — the distance-sense arm of the same “is this safe to ingest?” system as bitter/sour taste — with perfume-appreciation a softer overlay.
- How many odors humans can distinguish. “10,000” is folklore; the 2014 “one trillion” claim was reanalyzed and is not statistically supported; there is no defensible specific number.
- The strong “Proust” claim. That odor memories are more emotional/vivid has real support; that smell is uniquely “the memory sense,” or that odor memories are more accurate, runs past the evidence, especially versus other potent cues (music, faces).
- Human pheromones. The regressed, non-functional VNO and pseudogenized vomeronasal genes are solid; functional human pheromone communication is not established, and direct VNO-occlusion tests argue against a vomeronasal role.
And, as always, there is a great deal here we are sure of. The combinatorial logic of olfactory coding — one neuron, one receptor; broad tuning; odors as combinations; convergence onto glomeruli — is one of the most elegant and well-established stories in sensory neuroscience, and it explains, cleanly, how a few hundred detectors distinguish a practically limitless world of smells. You can rely on it.