Smell and Taste

The shortcut: why smell skips the thalamus

Trace any other sense from its receptor to your awareness and you will pass through the same building. Light hits the retina, and the signal travels to the lateral geniculate nucleus of the thalamus, then to visual cortex. Sound moves the cochlea, climbs through the brainstem, reaches the medial geniculate nucleus, then auditory cortex. A touch on the hand reaches the ventral posterolateral nucleus, then somatosensory cortex. Even taste, smell's chemical sibling, goes through the thalamus, arriving at the ventral posteromedial nucleus before it reaches the gustatory cortex.

Now trace smell. Odour molecules dissolve in the mucus of the olfactory epithelium, a postage-stamp of tissue in the roof of the nasal cavity. The receptor neurons there send their axons directly upward through the perforations of the cribriform plate, a sieve of bone in the base of the skull, and into the olfactory bulb sitting immediately above it. From the bulb, the mitral and tufted cells project along the olfactory tract to their targets. And those targets are the striking part.

Put those three destinations together and the phenomenology follows from the anatomy. A smell reaches the amygdala and the entorhinal cortex early, directly, and largely without the thalamic gate that filters, weights, and delays every other sense according to what the cortex expects. So when a particular smell retrieves a particular memory, it is not doing something mysterious. It is taking a road that was built for exactly that, and taking it fast.

Two caveats keep this honest. First, "no thalamic relay" means no obligatory one: there is a secondary olfactory route through the mediodorsal thalamic nucleus to the orbitofrontal cortex, and that pathway matters for conscious odour discrimination and for judging whether something smells good. What is unique to smell is that the primary route does not require it. Second, the claim that smells evoke memories with unusual emotional force is well supported experimentally, but researchers still debate exactly how much of that force comes from the anatomical shortcut and how much from the fact that odours are rarely rehearsed, so the memories they attach to are less worn down by retelling. The anatomy is certain. Its full explanatory weight is still being argued.

Four hundred receptors, and why that is enough

For most of the twentieth century nobody knew what an odorant receptor was. Vision had its rhodopsin, hearing had its hair cells, but the molecule that actually met a smell remained hypothetical. In 1991 Linda Buck and Richard Axel published the answer: a large multigene family, expressed in the olfactory epithelium, encoding a set of related G-protein-coupled receptors. The paper reshaped the field and won them the Nobel Prize in Physiology or Medicine in 2004.

Two features of the system they uncovered are worth pausing on, because together they solve a problem that looks unsolvable.

One neuron, one receptor. Each olfactory receptor neuron expresses, essentially, a single receptor type out of the human repertoire of roughly 400 functional ones. It is a specialist. And all the neurons expressing the same receptor, wherever they happen to sit in the epithelium, send their axons to the same one or two glomeruli, spherical tangles of neuropil, in the olfactory bulb. The bulb therefore holds an orderly map, not of space, but of receptor identity. Activate a given receptor and a given glomerulus lights up.

Many receptors, one odour. Here is the part that matters. A single odorant does not activate one receptor. It activates a subset, weakly here, strongly there, and a single receptor responds to many different odorants. The identity of an odour is therefore carried by the pattern of activation across the population, not by which single lock a single key opened.

Why build smell this way when vision and hearing are built differently? Because the stimulus space has a different shape, and this is the deep point. Light has a natural axis: wavelength. Order it and you get a spectrum, and three receptor types positioned along that axis suffice to specify a colour, because colour is a position on a line. Sound has a natural axis: frequency. Order it and you get a scale, and the cochlea is literally laid out along it, low frequencies at one end and high at the other.

Odour has no such axis. There is no physical dimension along which you can line up coffee, petrol, lavender, and burnt toast such that neighbours smell alike. Molecular weight will not do it. Neither will shape, nor volatility, nor any single chemical parameter: molecules that look almost identical can smell utterly different, and molecules with no structural resemblance can smell nearly the same. Smell space is high-dimensional and it is lumpy. A sense confronting a space like that cannot use the trick that vision and hearing use, which is to place a few receptors along a line and interpolate. It must instead sample the space with many broadly-tuned detectors and read the pattern. That is precisely what olfaction does, and the odd design turns out to be the correct design for the problem it faces.

The neuron that dies and is replaced

Neurons are, as a rule, cells you do not get back. That rule is the reason a stroke leaves a permanent deficit and a severed spinal cord does not reconnect. Olfactory receptor neurons are the loudest exception in the human body, and the reason they are an exception follows directly from where they sit.

An olfactory receptor neuron is a genuine neuron, with a cell body, dendrites, and an axon that runs into the brain. But one end of it is naked, in the mucus of the nasal cavity, exposed to the outside world. It breathes in whatever you breathe in: dust, viruses, solvents, smoke, cold dry air. No other neuron in your body is directly exposed to the environment like this, and the cost is that these neurons die, continuously, over weeks to months. If they were not replaced you would lose your sense of smell by early adulthood.

So they are replaced. A layer of basal stem cells sits beneath the receptor neurons in the epithelium, dividing and differentiating into new receptor neurons throughout life. Each new neuron must then grow an axon through the cribriform plate and find the correct glomerulus in the bulb, a remarkable feat of targeting performed continually in an adult brain that has otherwise finished wiring itself.

The turnover also explains a clinical fact. Smell loss after a viral infection is common and often recovers slowly, over months, because the epithelium must rebuild itself and rewire into the bulb. Smell loss after a head injury that shears the delicate axons at the cribriform plate is far less likely to recover, because the damage is to the road, not to the traveller.

Taste: five categories and a swallow-or-spit decision

Set beside smell's four hundred receptor types and combinatorial extravagance, taste looks impoverished. There are five well established basic tastes, and only five: sweet, sour, salty, bitter, and umami. Not five hundred. Five.

The obvious question is why. And the answer is not that evolution ran out of ideas. It is that the two chemical senses are answering different questions, and taste's question is much simpler.

Now look at what those five have in common. Every one of them is a verdict about a single decision that must be made in about a second, with the substance already in your mouth: swallow, or spit. Is there energy here? Is there protein? Is there salt? Is this rotten? Is this poison? Taste is not a sense for describing the world. It is a gatekeeper standing at the last checkpoint before your digestive tract, and a gatekeeper does not need vocabulary. It needs speed and it needs reliability. Five broad, unambiguous categories, each wired to an innate approach or withdrawal response, are exactly what that job requires; four hundred subtle discriminations would be a liability, because they would slow the decision and blur the alarm.

Smell, by contrast, works at a distance, before commitment, on objects that must be identified rather than merely accepted or rejected. Identification needs resolution. Rejection needs a klaxon. The two senses are built differently because they are doing different jobs, and once you see that, the asymmetry stops being a puzzle and starts being a design.

A note on the anatomy. Taste receptor cells sit in taste buds, clustered on the papillae of the tongue and also on the palate, epiglottis, and upper oesophagus. They are not neurons; they are specialised epithelial cells that synapse onto sensory nerve fibres. Three cranial nerves carry taste: the facial nerve from the front of the tongue, the glossopharyngeal from the back, and the vagus from the throat. All three converge in the nucleus of the solitary tract in the medulla, and from there taste ascends, through the thalamus, to the gustatory cortex in the insula. Taste, unlike smell, is a well-behaved sense that obeys the usual rules.

Flavour is not taste, and you can prove it in ten seconds

Everything so far sets up the payoff, which is the single most useful thing on this page.

When you eat a strawberry and describe its taste, almost nothing you are describing is taste. Your tongue is reporting five things: sweet, a little sour, and not much else. The strawberry-ness, the specific, unmistakable, complex identity of that fruit, is coming from your nose.

Here is how. Chewing warms and breaks up food, releasing volatile molecules. Those molecules do not stay in the mouth. They travel up the nasopharynx, the passage connecting the back of the throat to the nasal cavity, and reach the olfactory epithelium from behind, particularly on the exhale that follows a swallow. This is retronasal olfaction, smelling from the inside, and it is the dominant contributor to what you experience as taste.

The same demonstration is run on you every winter without your consent. When a cold blocks your nose, food goes flat and people say their sense of taste has gone. It has not. Their tongue is entirely intact and is still faithfully reporting sweet, sour, salty, bitter, and umami. What has gone is the retronasal channel, and with it roughly everything that made the food interesting. The complaint "I cannot taste anything" is, in almost every case, a complaint about smell.

Flavour, properly defined, is not a sense at all. It is a multisensory construction: taste from the tongue, retronasal odour from the nose, texture and temperature from mechanoreceptors and thermoreceptors in the mouth, and chemical irritation from the trigeminal nerve, which is what gives you the burn of chilli, the sting of carbonation, and the cooling of mint. None of those are tastes. Capsaicin does not activate a taste receptor; it activates a heat and pain channel, TRPV1, which is why chilli genuinely feels hot and why cold water fails to relieve it. Menthol activates TRPM8, the cold receptor, which is why mint feels cool. Your mouth is fooling you, and it is fooling you through the pain system.

The brain then binds all of this into a single, unified percept and locates it, wrongly, in the mouth. You experience strawberry as something happening on your tongue, when most of the information arrived at the roof of your nasal cavity. That mislocalisation is not a bug in your introspection. It is an example of the brain doing what it always does, which is to build the most useful interpretation of its inputs and present the interpretation, not the inputs. The same logic is set out for every sense in sensory processing.

When the chemical senses fail

Because smell has an unusual anatomy, it fails in unusual ways, and those failures are clinically informative.

Anosmia is the complete loss of smell; hyposmia is a reduction of it. The commonest causes are nasal obstruction, which simply prevents molecules reaching the epithelium, viral damage to the epithelium itself, and head injury, which can shear the fine axons where they pass through the cribriform plate. The first two often improve. The third frequently does not, and the mechanical reason is straightforward: the epithelium can rebuild, but the axons must then find their way back through a bone that may have scarred over.

Parosmia, in which familiar smells become distorted and often repellent, commonly appears during recovery from viral anosmia, and it has a plausible explanation in the biology set out above. As new receptor neurons regenerate, their axons must reconnect to the correct glomeruli in the bulb. If some of them connect to the wrong ones, the pattern of glomerular activation produced by a familiar odour is no longer the pattern the brain learned, and the odour is read as something else. A miswired combinatorial code produces a wrong answer, not a weak one. Coffee does not smell faint; it smells like burning rubbish.

Smell loss as an early sign. Reduced olfaction appears years before motor symptoms in Parkinson's disease and is common in early Alzheimer's disease. Why should a nose predict a brain disease? Because the olfactory system's central targets, including the entorhinal cortex, are among the earliest sites of pathology in Alzheimer's, and because the olfactory bulb is one of the first structures affected in Parkinson's. The nose is not predicting the disease; it is reporting damage that has already started in the structures it happens to project to. Smell testing is now a recognised, if non-specific, part of the early clinical picture in both conditions.

Ageusia, true loss of taste, is genuinely rare, and that rarity is itself instructive. Taste is carried by three separate cranial nerves on each side of the head, and taste buds are distributed across the tongue, palate, and throat. The system is heavily redundant, which is exactly what you would expect of a poison detector, because a poison detector that could be silenced by a single nerve injury would be a poor poison detector. Most patients who report losing their taste have lost their smell.

Three things everyone believes that are wrong

Related topics