Smell

Abstract

Smell is a type of sensation: the chemical sense that detects airborne molecules and converts them into the experience of odor. It begins at a few hundred kinds of odorant receptor in the nose, each tuned broadly rather than to a single molecule, so that an odorant is encoded not by one dedicated receptor but by the particular combination of receptors it activates. That combinatorial code, read out through the olfactory bulb and cortex, lets a modest receptor repertoire distinguish an enormous range of smells, and it is the organizing principle behind the molecular biology that earned a Nobel Prize. This article sets out the receptors and their code, how receptor activity becomes a perceived odor, how acute human smell is, and how the sense is measured, distinguishing smell as the sensory ability from olfactory perception, its downstream interpretation.

Keywords: smell, olfaction, odorant receptor, combinatorial code

A cup of coffee carried across a room announces itself before it is seen, and a single trace of smoke can empty a building. Yet the same sense runs so quietly beneath notice that most people rank it the most expendable of the senses, until a head cold flattens food to cardboard or a lost sense of smell turns a familiar kitchen into a place of faint unease. Smell is the oldest of the senses and, for a long time, the least understood: its receptors were not identified until 1991, and the scale of what the human nose can actually do is still argued over. The sense reveals its logic most clearly in its molecular machinery, where a few hundred broadly tuned receptors encode an odor by the pattern of their joint response rather than by any one of them acting alone (Buck & Axel, 1991; Malnic, Hirono, Sato, & Buck, 1999).

What Smell Is

Smell, or olfaction, is the sense that detects chemicals carried in the air and renders them as the conscious experience of odor. It is one of the two chemical senses, the other being taste, and it is defined by MeSH as the ability to detect scents, the function served by the olfactory receptor neurons of the nose. As a sensory ability it sits alongside vision, hearing, touch and taste as a channel through which the nervous system registers the world; where this article treats smell as that sensory channel, the richer business of recognizing, naming and remembering particular odors is the domain of olfactory perception.

The sense begins in a patch of specialized epithelium high in the nasal cavity, where millions of olfactory sensory neurons extend fine cilia into the overlying mucus. Dissolved odorant molecules bind to receptor proteins on those cilia, and a neuron that is sufficiently excited sends a signal along its axon to the olfactory bulb of the brain. What makes the system remarkable is not the number of neurons but the economy of its coding: a few hundred kinds of receptor, each responding to a range of molecules, together generate a space of possible response patterns vast enough to represent far more distinct odors than there are receptors (Firestein, 2001). The sections that follow build the sense from that molecular layer upward: first the receptors and their combinatorial code, then the route from receptor activity to a perceived odor, then the question of how much the human nose can actually resolve, and finally how the sense is measured.

The Odorant Receptors and the Combinatorial Code

For most of the twentieth century the receptors of smell were entirely unknown, a gap that set olfaction apart from vision and hearing, whose receptor mechanisms were understood in detail. The gap closed in 1991, when Buck and Axel identified a large multigene family encoding the odorant receptors, proteins of the G-protein-coupled receptor class expressed on the sensory neurons of the nose. The family is the largest in the mammalian genome, numbering around a thousand genes in rodents and about four hundred functional genes in humans, and its discovery provided the molecular basis of odor recognition that the field had lacked, work for which the two shared the 2004 Nobel Prize in Physiology or Medicine (Buck & Axel, 1991).

Two organizing principles followed. The first is that each sensory neuron expresses just one of the hundreds of receptor genes, and all neurons expressing a given receptor send their axons to a small number of shared targets, the glomeruli, in the olfactory bulb. Mombaerts and colleagues made this convergence visible by genetically labelling the neurons carrying one receptor and watching their axons gather at the same two glomeruli, so that the receptor a neuron chooses determines where in the bulb it reports, converting the molecular identity of the receptor into a fixed spatial address (Mombaerts et al., 1996). The second principle is the combinatorial code. Malnic and colleagues exposed individual sensory neurons to a panel of odorants and found that one receptor responds to several different molecules and one molecule activates several different receptors, so that each odorant is represented by a particular combination of receptors rather than by a dedicated one. Because combinations grow far faster than their elements, a few hundred receptors can encode a practically unlimited number of odors (Malnic, Hirono, Sato, & Buck, 1999). Figure 1 shows the principle: one odorant lights a subset of receptors, and the pattern of lit receptors is the code for that odorant.

Figure 1

The Combinatorial Receptor Code for Odor

The power of the scheme is purely combinatorial. If each of a set of receptors either does or does not respond to a molecule, then the number of distinct on-or-off patterns doubles with every receptor added, so even ten receptors yield more than a thousand possible codes and a few hundred yield a number with no practical ceiling. The first demonstration lets the reader switch odorants and watch the receptor pattern and its barcode change, and counts the codes that a given number of receptors can represent.

From Receptors to Perception

The receptor code is only the first step. The axons of the sensory neurons deliver the code to the glomeruli of the olfactory bulb, where it becomes a spatial pattern of activity across the bulb surface, and from there the signal travels to the olfactory cortex and on to regions concerned with memory and emotion. Firestein set out how the system makes sense of this input: the broadly tuned receptors, the convergence onto glomeruli, and the downstream circuitry together transform a molecular event at the nose into a neural representation that the brain can read as a particular smell (Firestein, 2001). Unlike the senses of sight and hearing, whose primary cortical areas lie far from the receptors, the olfactory pathway reaches cortex in remarkably few synapses and bypasses the thalamic relay that the other senses pass through, which is often invoked to explain the close tie between smell, memory and emotion.

A central idea in the perception of smell is the odor object. The brain does not experience a list of activated receptors; it experiences coffee, or smoke, or a rose, as a unified thing. Gottfried reviewed the central mechanisms by which the olfactory system binds the combinatorial input into such objects, a synthetic process in which the whole percept is more than the sum of its molecular parts and in which learning and context shape what is perceived (Gottfried, 2010). Yeshurun and Sobel pressed the point from the psychological side, arguing that human olfactory perception is organized around a single dominant dimension, roughly how pleasant or unpleasant a smell is, and that people are strikingly poor at naming odors even when they recognize them vividly, so that an odor is, in their phrase, not worth a thousand words (Yeshurun & Sobel, 2010). Table 1 sets out the stages from molecule to perceived odor.

How Acute Is Human Smell?

A persistent piece of folklore holds that humans are poor smellers, dwarfed by dogs and rodents, a notion that traces to nineteenth-century anatomists who contrasted the small human olfactory bulb with the large brain above it. McGann traced this claim to its origin and showed it to be a myth: human olfactory performance, measured carefully, is excellent, comparable to other mammals on many odors and superior on some, and the small relative size of the human bulb is a poor guide to its absolute capacity (McGann, 2017). Shepherd had earlier made the same argument on physiological and evolutionary grounds, pointing out that the human brain devotes substantial cortical processing to smell and that the sense is far more important to human behavior than the folklore allows (Shepherd, 2004).

The reaction against the myth produced its own overstatement. Bushdid and colleagues estimated, from how well people discriminated mixtures of many odorous molecules, that humans can distinguish well over a trillion olfactory stimuli, a figure that spread quickly as a striking upper bound on the sense (Bushdid, Magnasco, Vosshall, & Keller, 2014). The number rests on a combinatorial extrapolation from a limited sample of mixtures, and that method was criticized as mathematically fragile, so the precise figure should be read as a vivid illustration rather than a measured constant; the robust conclusion is that the discriminable space is very large, not that it has any particular size. Individual noses also differ genetically: Trimmer and colleagues showed that common variation across the odorant receptor genes changes how specific odors are perceived, so that two people with different receptor repertoires can genuinely smell the same molecule differently (Trimmer et al., 2019). The second demonstration makes the combinatorics concrete, letting the reader vary the size of an odorant palette and the number of components in a mixture and watch how fast the number of possible mixtures grows.

Measuring the Sense

Because odor experience is private, measuring smell requires tasks that externalize it, and two kinds dominate. An identification test presents a set of common odors and asks the person to name each from a short list of choices; a threshold test presents an odor at a series of concentrations and finds the weakest one reliably detected. Doty and colleagues built the standard identification instrument, the University of Pennsylvania Smell Identification Test, which seals each odor in a scratch-and-sniff microcapsule and scores the number correctly identified out of forty, giving a reliable, norm-referenced measure of olfactory function that is still in routine clinical use (Doty, Shaman, & Dann, 1984). Threshold testing complements it by locating the detection limit on a concentration scale, the point at which the psychometric function crosses the halfway mark between chance and certainty.

Psychophysics also serves as a testbed for theories of how smell works at all. A long-running debate asks whether receptors recognize a molecule by its shape or by its molecular vibrations; Keller and Vosshall put the vibration theory to a direct psychophysical test, asking whether people could smell the difference between a molecule and its isotope-substituted twin, which shares shape but differs in vibration, and found no support for the prediction the theory required, a result that kept the shape-based account as the working model (Keller & Vosshall, 2004). The third demonstration runs a detection-threshold task, letting the reader set the concentration of an odor and read off the probability of detection as the psychometric function predicts.

Worked Example

The central facts of smell become concrete when the numbers are worked, and the demonstrations reproduce this arithmetic. Take the combinatorial code first. If each receptor type either responds to a molecule or does not, a set of receptors acts like a row of binary switches, and the number of distinct on-or-off patterns is two raised to the number of receptors. Ten receptors therefore yield two to the tenth, which is 1,024 distinct codes; twenty yield 1,048,576. The human repertoire of about four hundred functional receptors makes the code space astronomically large, which is why a few hundred receptors suffice to represent far more odors than there are receptor types. This is the count the first demonstration displays as receptors are added.

Now the discriminable space estimated from mixtures. The number of distinct mixtures of k components that can be drawn from a palette of n odorants is the binomial coefficient n-choose-k. From a palette of 10, the three-component mixtures number 10-choose-3, which is 120; from a palette of 20, the five-component mixtures number 20-choose-5, which is 15,504. Scaled up to the palette used in the trillion-smells study, 128 odorants taken 30 at a time gives a number on the order of ten to the twenty-ninth, far larger than any estimate of what can actually be discriminated. The gap is the whole point of the critique: the count of possible mixtures is exact arithmetic, but the fraction of them a person can actually tell apart is an extrapolation, and it is the extrapolation, not the arithmetic, that is uncertain. The second demonstration computes n-choose-k as the palette and mixture size are varied.

Finally the detection threshold. Model the probability of detecting an odor as a logistic function of log concentration, p equals one divided by one plus e to the minus (x minus m) over s, where x is the log concentration, m is the threshold and s sets the steepness. With the threshold m at zero and s equal to one half, a concentration one log unit below threshold gives p equal to about 0.12, at threshold p is exactly 0.5, and one log unit above threshold p is about 0.88. The halfway point of the curve defines the measured threshold, and each log unit on the axis is a tenfold change in concentration. The third demonstration plots this curve and reads the detection probability at the concentration the reader sets.

Discussion

Smell began the modern era as the sense whose receptors were unknown and ends it as a worked example of combinatorial coding. The discovery of the odorant receptor family resolved the central mystery, and the two principles that followed, one receptor per neuron with convergence onto shared glomeruli, and one odor encoded by a combination of receptors, together explain how a few hundred receptor types represent an effectively unlimited range of odors (Buck & Axel, 1991; Mombaerts et al., 1996; Malnic, Hirono, Sato, & Buck, 1999). The route from that code to a perceived odor object, assembled in cortex and shaped by learning, is the part still least understood, and it is where the psychology of smell meets its neuroscience (Firestein, 2001; Gottfried, 2010).

The study of smell also offers a lesson in how a field corrects itself. The myth of poor human olfaction, built on a misreading of bulb size, was overturned by careful measurement, and the vivid counter-claim of a trillion discriminable smells was in turn tempered by scrutiny of its mathematics (Shepherd, 2004; McGann, 2017; Bushdid, Magnasco, Vosshall, & Keller, 2014). What survives both corrections is a sense that is acute, highly individual, and far more cognitively central than its reputation suggested, with genetic differences in the receptor repertoire giving each person a partly private olfactory world (Trimmer et al., 2019).

Current Directions

Two developments have reshaped the study of smell in recent years. The first is structural. For three decades the odorant receptors were known by their genes but not seen in action, because this class of receptor is notoriously hard to crystallize. Billesbolle and colleagues solved the first cryo-electron-microscopy structure of a human odorant receptor bound to an odorant, showing exactly how a smell molecule sits in the receptor pocket and begins the signal, and providing the template for understanding how the broadly tuned receptors achieve their characteristic breadth (Billesbolle et al., 2023). The structure turns the combinatorial code from an inference about response patterns into a problem that can be studied at the level of atoms.

The second development came from the clinic. When COVID-19 spread in 2020, a sudden anosmia emerged as one of its most distinctive early signs, and a global consortium documented that the disease is associated with severe impairment of smell, taste and chemesthesis, far beyond ordinary nasal congestion (Parma et al., 2020). The episode brought olfactory loss to wide public attention, drove a surge of research into how the virus damages the olfactory system and how the sense recovers, and underlined a point the field had long made quietly: that anosmia is not a trivial inconvenience but a genuine impairment of a sense woven into eating, memory and emotion.

Common Misconceptions

Humans are poor smellers compared with other animals.

This is a nineteenth-century myth built on the small relative size of the human olfactory bulb. Careful measurement shows human olfactory discrimination is excellent, comparable to other mammals on many odors and better on some (McGann, 2017; Shepherd, 2004).

Each smell has its own dedicated receptor.

There are far too few receptor types for that. One receptor responds to many molecules and one molecule activates many receptors, so an odor is encoded by a combination of receptors, not a single labelled line (Malnic, Hirono, Sato, & Buck, 1999).

Science has established that humans can smell exactly a trillion odors.

The trillion figure is an extrapolation from a limited sample of odor mixtures, and its mathematics was criticized. The defensible claim is that the discriminable space is very large, not that it has any exact size (Bushdid, Magnasco, Vosshall, & Keller, 2014).

Smell and olfactory perception are the same thing.

Smell is the sensory ability to detect odorants; olfactory perception is the downstream recognition, naming and interpretation of odors. The two are closely linked but are treated as distinct, as the odor-object literature makes clear (Gottfried, 2010).

Glossary

Anosmia.

The loss of the sense of smell, whether partial or total; its sudden onset became a widely recognized early sign of COVID-19.

Chemical sense.

A sense that detects molecules rather than energy; smell and taste are the two chemical senses, smell responding to airborne odorants.

Combinatorial code.

The representation of an odor by the particular combination of receptors it activates, rather than by a single dedicated receptor, so that a few hundred receptors can encode an enormous range of odors.

Detection threshold.

The weakest concentration of an odor that a person can reliably detect, located on a concentration scale at the point where the psychometric function crosses the halfway mark between chance and certainty.

Glomerulus.

A spherical cluster of synapses in the olfactory bulb where the axons of all sensory neurons expressing the same receptor converge, giving each receptor type a fixed spatial address.

Odor object.

The unified percept of a smell, such as coffee or rose, that the brain constructs from the combinatorial receptor pattern, shaped by learning and context rather than read off molecule by molecule.

Odorant receptor.

A G-protein-coupled receptor protein on the cilia of an olfactory sensory neuron that binds odorant molecules; humans express about four hundred functional types, the largest gene family in the genome.

Olfactory bulb.

The first brain structure of the olfactory pathway, where sensory-neuron axons synapse in glomeruli and the receptor code becomes a spatial map of activity.

Olfactory cortex.

The cortical region that receives projections from the olfactory bulb and assembles the receptor code into a recognizable odor, reached in remarkably few synapses and without the thalamic relay the other senses pass through.

Olfactory perception.

The downstream recognition, naming and interpretation of odors, as distinct from smell as the sensory ability to detect them.

Olfactory sensory neuron.

A receptor neuron in the nasal epithelium that expresses a single odorant receptor type and sends its axon directly to the olfactory bulb.

Psychometric function.

The curve relating the probability of detecting a stimulus to its intensity; its halfway point between chance and certainty defines the detection threshold.

Sensation.

The registration of physical stimuli by the sensory receptors, the broad category under which smell is classified as a chemical sense.

Smell identification test.

A standardized clinical instrument that presents a set of common odors to be named from short lists of choices, scoring olfactory function against population norms.

Key Researchers

Richard Axel

(b. 1946). Columbia University (Zuckerman Institute); co-discoverer of the odorant receptor gene family and, with Mombaerts, of how receptor identity organizes the olfactory sensory map onto the bulb. He shared the 2004 Nobel Prize in Physiology or Medicine. Faculty Page - Wikipedia - Wikidata

Linda B. Buck

(b. 1947). Fred Hutchinson Cancer Center; with Richard Axel she discovered the odorant receptor gene family in 1991, showing that each sensory neuron expresses one receptor type and that odor identity is read from the combination of receptors activated. She shared the 2004 Nobel Prize in Physiology or Medicine for the work. ORCID - Faculty Page - Wikipedia - Wikidata

Stuart Firestein

. Columbia University; studies the physiology of the olfactory receptor neuron and the transduction of odor into neural signals, and his widely cited synthesis set out how the olfactory system builds odor percepts from receptor activity. ORCID - Faculty Page - Google Scholar - Wikipedia - Wikidata

Gordon M. Shepherd

(1933-2022). Yale University School of Medicine; pioneered the study of olfactory bulb microcircuitry and the concept of olfactory images, and argued that human olfaction is far more acute than the myth of a poor human sense of smell allowed. Wikipedia - Wikidata

Noam Sobel

. Weizmann Institute of Science; leads psychophysical and neuroimaging work on human olfaction, including how odors are represented as perceptual objects and how smell carries social and emotional information. ORCID - Faculty Page - Google Scholar

Leslie B. Vosshall

(b. 1965). The Rockefeller University (Howard Hughes Medical Institute); studies the molecular neurobiology of olfaction and olfactory coding, and her psychophysical work argued that humans can discriminate an enormous range of olfactory stimuli, challenging the textbook estimate. Faculty Page - Google Scholar - Wikipedia

Frequently Asked Questions

What is smell?

Smell, or olfaction, is the chemical sense that detects airborne molecules and turns them into the experience of odor. It begins when odorant molecules bind receptors on the sensory neurons of the nose, and it is one of the two chemical senses, the other being taste (Firestein, 2001).

How many odorant receptors do humans have?

Humans express roughly four hundred kinds of functional odorant receptor, part of the largest gene family in the genome. Each responds to a range of molecules rather than to one, which is what makes the combinatorial code possible (Buck & Axel, 1991).

How can so few receptors detect so many smells?

Through a combinatorial code. One receptor responds to many molecules and one molecule activates many receptors, so each odor is represented by a pattern of receptor activity. Because combinations grow far faster than their elements, a few hundred receptors can encode a practically unlimited number of odors (Malnic, Hirono, Sato, & Buck, 1999).

Is the human sense of smell really poor?

No. The idea that humans are poor smellers is a nineteenth-century myth based on the small relative size of the human olfactory bulb. Measured carefully, human olfactory discrimination is excellent and in some cases exceeds that of animals long assumed to be superior (McGann, 2017).

Can humans really smell a trillion different odors?

The trillion figure is an extrapolation from how people discriminated a limited sample of odor mixtures, and its mathematics was criticized. The safe conclusion is that the number of discriminable smells is very large, not that it is any exact value (Bushdid, Magnasco, Vosshall, & Keller, 2014).

Why does smell trigger memories and emotions so strongly?

The olfactory pathway reaches the cortex in very few synapses and bypasses the thalamic relay the other senses pass through, connecting closely with regions concerned with memory and emotion. The brain also binds odors into unified objects shaped by past experience (Gottfried, 2010).

How is the sense of smell measured?

Chiefly by identification tests, which ask a person to name common odors from short lists, and by threshold tests, which find the weakest concentration reliably detected. The University of Pennsylvania Smell Identification Test is a widely used standardized identification instrument (Doty, Shaman, & Dann, 1984).

Why did COVID-19 cause loss of smell?

A sudden loss of smell emerged as a distinctive early sign of COVID-19, and a global study documented severe impairment of smell, taste and chemesthesis in the disease, beyond ordinary congestion. The finding drove new research into how the virus damages the olfactory system and how the sense recovers (Parma et al., 2020).

References