A woman in London, a former sommelier, who has not tasted wine properly since March 2021. Every glass, regardless of grape or vintage, arrives on her palate as a wash of acetone and burnt rubber. Her Burgundy smells like a garage. Her Sancerre, like nail polish remover. She has not lost her sense of smell. A worse fate has happened: her brain has begun rewriting it.
11 min read
Her condition is called parosmia, the systematic distortion of real odours into phantom ones, typically grotesque. It is the brain's cruel parody of perception: everything registers, but nothing registers correctly. Coffee smells like sewage. Roses smell like chemicals. The body of a lover smells like rotting meat. The world becomes a funhouse mirror built entirely of scent.
But parosmia has a stranger, quieter cousin. Phantosmia, from the Greek phantasma, an apparition, is the perception of smells that have no source whatsoever. No molecule has entered the nose. No receptor has fired in response to the external world. And yet the smell is there, vivid and insistent: burnt toast at three in the morning in a clean kitchen, cigarette smoke in an empty room, the sweet chemical tang of something rotting in a house where nothing has died.
These are not metaphors. They are neurological events. And they reveal an unsettling truth about perception itself, one that perfumery, perhaps more than any other art, is uniquely positioned to understand.
The olfactory bulb's peculiar position in the brain
To grasp what phantosmia discloses, one must first understand the olfactory bulb, and why it occupies such a peculiar position in the architecture of the human brain.
Every other sensory system is insulated. Vision passes through the retina, the optic nerve, the lateral geniculate nucleus, the thalamus, before arriving at the visual cortex, a relay chain of astonishing length, each station filtering and interpreting before the signal moves on. Sound travels a similarly bureaucratic route: cochlea, auditory nerve, brainstem nuclei, medial geniculate body, thalamus, auditory cortex. Touch, taste, proprioception, all are routed through the thalamus, that great switchboard operator of the brain, which decides what reaches consciousness and in what order.
Smell refuses this arrangement entirely. The olfactory bulb, a pair of structures no larger than blueberries, sitting at the base of the frontal lobe, receives input directly from the olfactory epithelium, a postage-stamp-sized patch of tissue high in the nasal cavity. Between the outside world and the brain, there is almost nothing: a thin lattice of bone called the cribriform plate, perforated like a sieve, through which the axons of olfactory receptor neurons thread themselves directly into the bulb. No other part of the central nervous system is so nakedly exposed to the environment. The olfactory bulb is the brain's open window, or, more accurately, its open wound.
From the bulb, signals travel not to the thalamus but directly to the piriform cortex and the amygdala, the seat of emotional memory. This is why smell is so often described as the most primitive sense, the most emotionally saturated, the most resistant to language. Smell is not primitive. It bypasses the editorial machinery that other senses must pass through. This architectural shortcut is also what makes olfactory fatigue so ruthlessly efficient at erasing constant stimuli from awareness. A smell enters consciousness raw, unmediated, already entangled with memory and feeling before the prefrontal cortex has had time to form a thought about it.
This architecture explains a great deal. It explains why the smell of a particular laundry detergent can reduce an adult to tears without warning. It explains why olfactory memory is so durable and so resistant to voluntary recall, you cannot summon a smell the way you can summon a melody, but when the smell hits unbidden, the memory it carries is total. And it explains, critically, why damage to the olfactory system produces such bizarre and specific forms of suffering.
How SARS-CoV-2 attacked the olfactory epithelium
The SARS-CoV-2 virus, it turned out, had a particular affinity for the olfactory epithelium, as demonstrated in a 2020 study by Brann, Tsukahara, and colleagues at Harvard Medical School, published in Science Advances. The receptor it exploited for cellular entry, ACE2, was expressed in high concentrations on the sustentacular cells that support olfactory neurons. The virus did not need to reach the brain to destroy smell. It needed only to reach the nose.
The scale of the resulting olfactory damage was unprecedented in modern neurology. Estimates vary, but a 2022 meta-analysis published in the British Medical Journal covering over 600,000 patients suggested that between forty and sixty-five percent of Covid-19 patients experienced some degree of olfactory dysfunction. For most, it resolved within weeks. For millions, it did not. By late 2021, clinics specialising in post-viral anosmia, previously a niche subspecialty, were overwhelmed. A generation of people discovered, for the first time, what it means to live in a world without smell.
But the loss of smell was only the first act. As damaged olfactory neurons began to regenerate, a process unique to the olfactory system, which retains neuroplastic capacity throughout life, many patients found that their smell did not return cleanly. It returned wrong. Parosmia set in: the distortions, the grotesque substitutions, the sense that the world had been rebuilt slightly off-axis. And for a smaller but significant subset, phantosmia emerged alongside it, smells conjured from nothing, generated entirely within the brain.
The neuroscience of this is both well-understood and deeply strange. When olfactory receptor neurons are destroyed and begin to regrow, they must find their way back to the correct glomeruli in the olfactory bulb, the precise docking stations where specific receptor types converge. This process is not always accurate. Axons miswire. Receptors connect to the wrong glomeruli. The result is a scrambled map: the brain receives signals that are structurally coherent but informationally wrong, like a piano whose strings have been reattached to the wrong hammers. Press middle C and you get F-sharp. Press F-sharp and you get something that is not quite any note at all.
Phantosmia goes further. In phantosmia, the brain is not misinterpreting a signal. It is generating one. The olfactory cortex, deprived of its normal input, or receiving only garbled fragments, begins to fill in the gaps. It composes. It invents. It produces olfactory percepts that are often startlingly specific: not vague impressions of "something burning" but the precise, unmistakable smell of burnt toast, or gasoline, or a particular brand of cigarette the patient has not encountered in decades. The brain, left alone in a dark room, starts talking to itself. And what it says is detailed, coherent, and entirely fictional.
Phantom smells predate Covid by centuries
This phenomenon is not new. It merely lacked, until Covid, a sufficiently large population of sufferers to force it into public consciousness.
Fyodor Dostoevsky, who suffered from temporal lobe epilepsy throughout his adult life, as documented by the neurologist and medical historian John R. Hughes in a 2005 review in Epilepsy and Behavior, described the auras preceding his seizures in terms that neurologists now recognise as phantosmic episodes. Before the convulsion arrived, before the terror and the loss of consciousness, there was a moment of strange, overwhelming beauty. He described it to his friend Strakhov as an experience of piercing clarity, a sensation of the brain operating at some higher frequency, accompanied by what witnesses described as his sudden, beatific stillness. Temporal lobe epilepsy is well documented to produce olfactory hallucinations during auras, and Dostoevsky.s episodes were no exception to the broader pattern. Impressions of scent that seemed to arrive from nowhere and everywhere simultaneously, carrying with them a conviction of absolute meaning that dissolved the instant the seizure began.
Temporal lobe epilepsy has long been associated with olfactory hallucinations. The uncinate fasciculus, a white-matter tract connecting the temporal lobe to the orbitofrontal cortex, passes through regions intimately involved in olfactory processing. When seizure activity spreads through these circuits, the result is often a sudden, vivid, involuntary smell. Patients describe it variously: a smell of burning, of rubber, of flowers, of something indescribable but intensely familiar. The phenomenon is called an uncinate fit, and it has been documented since John Hughlings Jackson, the father of English neurology, first described it in his clinical reports at the National Hospital for the Paralysed and Epileptic in the 1880s. It is, in essence, the brain's olfactory system firing without permission, a rogue composition, generated from within.
What links the epileptic's phantom smell to the Covid patient's distorted coffee is a single principle, and it is the same principle that makes phantosmia so philosophically unsettling: the brain does not passively receive olfactory information. It actively constructs it. Smell is not a recording. It is a performance.
What happens when you smell a rose, at molecular level
The implications of this are worth sitting with.
When you smell a rose, what is happening, at the most granular level, is that a cloud of volatile molecules (several hundred distinct compounds, in the case of a centifolia rose, as catalogued by researchers at the INRA in France) is binding to a subset of your roughly four hundred types of olfactory receptors. Each molecule activates a different combination of receptors. The pattern of activation is transmitted to the olfactory bulb, where it is processed into what neuroscientists call an "odour object", a unified percept that the brain recognises as "rose." But this odour object is not a photograph of the molecular reality. It is a construction, a model, built by the brain from fragmentary chemical data and shaped by memory, expectation, context, emotional state, and genetic variation in receptor expression.
Two people smelling the same rose are, in a meaningful neurological sense, smelling different things. Not because the molecules differ, but because the brains composing the percept differ. The receptor repertoire is not identical across individuals, genetic polymorphisms in olfactory receptor genes mean that some people are functionally anosmic to specific molecules that others find overpowering. The emotional associations are not identical. The memories triggered are not identical. The rose is the same. The experience of the rose is irreducibly personal.
Phantosmia merely makes visible what is always true: that the brain is the composer, not the audience. In normal olfactory perception, the brain composes in response to molecular input, it has a score to follow, however loosely. In phantosmia, the brain composes without a score. The orchestra plays on, but the sheet music is blank. And the unsettling thing, the thing that should give us pause, is that the resulting performance is often indistinguishable, from the inside, from the real thing. The phantom smell of burnt toast is not experienced as a hallucination. It is experienced as burnt toast. The brain's composition is so convincing that consciousness cannot tell the difference.
This is not a bug in the system. It is the system. Perception has always been a creative act. The brain has always been generating its world as much as receiving it. We know this from visual neuroscience, the blind spot, change blindness, the McGurk effect, but olfaction makes the point with a particular, uncomfortable clarity, because smell is the sense we trust most instinctively and interrogate least. We doubt our eyes. We question our ears. We almost never question our nose.
Smell training and the Hummel protocol
The treatment for phantosmia and parosmia is as low-tech as it is effective. It is called smell training, and its most widely validated protocol was developed by Thomas Hummel at the Smell and Taste Clinic of the Technical University of Dresden. The method is simple to the point of absurdity: the patient sniffs four specific odours, rose, eucalyptus, lemon, and clove, twice daily, for at least twelve weeks. Each sniff lasts ten to twenty seconds. The patient is instructed to concentrate, to attempt to recall what the smell should be, to engage memory and attention simultaneously with the physical act of inhalation.
It works. Not for everyone, not completely, but with a consistency that Hummel and his colleagues have demonstrated in multiple controlled trials, including a key 2009 study published in The Laryngoscope. Patients who undergo structured smell training show measurably greater olfactory recovery than those who do not. The mechanism is neuroplasticity: the deliberate, repeated activation of olfactory circuits guides regenerating neurons toward their correct targets, reinforces weakened synaptic connections, and, critically, retrains the brain's predictive models of what a given pattern of receptor activation should mean. You are not merely exposing the nose to a stimulus. You are teaching the brain to compose correctly again.
The choice of the four odours is not arbitrary. Rose, eucalyptus, lemon, and clove were selected because they represent four primary odour categories, flowery, resinous, fruity, and spicy, providing broad coverage of the receptor repertoire. They are also, and this matters, culturally familiar: the brain's predictive model has strong priors for these smells, making the retraining process more efficient. Familiarity is not incidental to the treatment. It is the treatment. The brain heals faster when it knows what it is supposed to be hearing.
The parallel to musical training is not incidental either. A pianist recovering from a hand injury does not begin with Rachmaninoff. She begins with scales, simple, repetitive, structurally fundamental patterns that re-establish the neural pathways underlying more complex performance. Smell training is olfactory scales. It is the brain relearning its own instrument.
The perceiver is a collaborator, not a recorder
For those who work with fragrance, who spend their lives manipulating the materials of olfactory experience, phantosmia is a revelation. It confirms what the practice of perfumery has always implicitly known: that smell is not a passive sense but a creative one. That the perceiver is not a recording device but a collaborator. That the space between a molecule and a memory is not empty, it is filled with the brain's own compositional intelligence.
A perfumer, constructing an accord, is not assembling a stimulus. She is writing a score that another brain will perform. The performance will never be identical to the score. It cannot be. The performer, the wearer, the smeller, brings to the encounter a lifetime of olfactory experience, a unique receptor genotype, an emotional history that no other human shares. The smell of a perfume on skin is not a fact. It is an event, a collaboration between the composition and the consciousness that receives it.
Phantosmia simply reveals what happens when the collaboration breaks down, when the wearer's brain begins to improvise without the composer's input. The phantom smells are the brain's own perfumes, crude, often unpleasant, but structurally genuine olfactory experiences, generated by the same neural machinery that produces the experience of a tuberose absolute or a bergamot expression. They are proof that the machinery of smell is fundamentally generative. It does not need the world to create.
This is not a comfortable thought. We prefer to believe that our senses give us the world as it is, that perception is a window, not a painting. But the olfactory system, with its direct neural exposure, its bypassing of the thalamic checkpoint, its intimate entanglement with emotion and memory, has always been the sense that most openly refuses this fantasy. Smell has always been constructed. Smell has always been personal. Smell has always been, in the deepest neurological sense, a creative act.
The millions of people who lost their smell to a virus and found, in its place, a distorted or phantom world, learned this the hard way. The brain is not a microphone, faithfully recording the chemical environment. It is an orchestra, one that plays from a score when the score is available, and improvises when it is not. The music never stops. The question is only whether the composition reflects the world outside or the world within.
Precisely because the brain is a composer, the quality of the score matters immensely. A great perfume does not override the brain's compositional intelligence. It engages it. It provides a structure rich enough and complex enough to sustain the brain's own creative interpretation, the way a great piece of music provides a framework within which every performance is unique.
The phantom smell of burnt toast is the brain composing alone, with no score, from fragments and noise. A perfume is the opposite: a score so detailed, so considered, so materially grounded that the brain's performance of it becomes richer than either the composition or the consciousness could produce alone.
That collaboration, between molecule and memory, between the external world and the brain's own compositional intelligence, is what we mean when we say someone is wearing a perfume. Not applying it. Not passively receiving it. Wearing it: an active, creative, irreducibly personal act of perception.
The orchestra is always playing. The question is what you give it to perform.
