Does a dog hear worse when he's sniffing?
- Paws To Peaks
- Jun 8
- 32 min read
Three seconds during which a dog stops breathing. Switching attention between the senses — and reading a dog's behaviour.
Equipment and circumstances
I bought a small camera to record how Mrok interacts with other dogs and how he handles unusual circumstances — in a kayak, for example. Observing from the outside isn't always enough; I wanted to be able to re-watch those moments later, calmly, and catch details you don't pick up in the field.
The footage I'm writing about came about essentially by accident. I thought it would be interesting to see the world from Mrok's perspective, so I attached the camera to the upper part of his collar — the same one that already holds his GPS tracker and VHF radio. The whole setup looks a bit sci-fi, but he doesn't mind. The wide-angle lens sits a bit above Mrok's eye line when he's standing with his head in a neutral position; when he lowers his head to work with his nose, the frame drops even further. That's why you get those sliding shots of grass and sky — the lens position depends on what the dog is doing.
The camera is a DJI Osmo Nano. Its microphone is good enough that you can hear Mrok's breathing clearly. That turned out to be more important than I thought when I bought it.
I rewatched the recordings several times. Each pass revealed more details and prompted... another rewind.
That, essentially, is the starting point of this article.
What this article isn't
I don't work professionally in medicine, neurology, or animal behaviour. I'm Mrok's guardian, who started watching the footage from a collar camera, saw something that intrigued him, and went looking for explanations. I like to observe, learn, and draw conclusions. The footage of Mrok's exploration prompted me to dig into a substantial body of work written by top-class specialists, along with the commentary on it, across a wide spectrum of fields. Anyone can find these materials and analyse them for themselves — I list the key works at the end.
Please take this article with a grain of salt — more as the work of an enthusiast than a specialist. That doesn't change the fact that the source materials are knowledge in its purest form. If, through this article, I help someone look at their own dog differently, with more understanding, then it has served its purpose.
Mrok, the camera, the walk, and a chance decision
On one of our walks, I thought it would be fun to see the world from Mrok's perspective. We don't usually interfere with his exploration, especially since we know he's continuously keeping track of where we are. I was curious whether the footage would show him turning his head to look for us, or whether he always positions himself so we're inside his wide field of view — even when it looks like he isn't paying us any attention at all.
Watching the footage back several times in a row, sometimes considerably slowed down, I saw something that surprised me. Mrok was walking with his nose to the ground, working in full scent mode. You can hear that characteristic, fast sniffing rhythm in the footage — the camera's microphone picks it up clearly. Suddenly the rhythm stops. The breath goes quiet. The head lifts. The image stops swaying. And you can see, after a slight correction in the frame's axis, that Mrok has turned his head in a specific direction, setting his ears at a precise angle (ACDs have upright ears, so it's visible). Three seconds of stillness. Then his head drops back down, sniffing picks up in the same rhythm as before, as if nothing had happened.
I scrubbed back through the footage. The same sequence repeated more than a dozen times. Each time the sniffing clearly weakened or stopped. In another scene, when Mrok spotted a dog on the horizon, the same pattern again — only this time vision dominated; the nose stopped working for a few seconds, the image stabilised on the moving silhouette in the distance, Mrok's body visibly stiffened. When the unfamiliar dog disappeared behind a bush, smell became the dominant sense again. The modality switching — the fluid transition between the senses — was surprising. Next scene: Mrok suddenly stops and angles his head oddly to the side. As I figured out later, he was acoustically locating something I couldn't see. Before that, for a good ten seconds he had been walking with his nose to the ground and, going by the footage, paying that "something" no attention at all.
If you asked an average dog person what was going on in those scenes, you'd hear: "Well, he heard something. Saw something. Nothing unusual." It's true, but it's like saying a plane "flies because it has wings." It doesn't describe the mechanism. Because if Mrok just "heard" or "saw" something, he'd be doing it constantly — including while he was sniffing. And he isn't. His nervous system manages access to his senses in a far more refined way than intuition would suggest. Those three seconds of silence, what precedes them, and what follows, are a window onto one of the oldest mechanisms of the mammalian brain.
How to read this article
I've broken the recorded scenes into four phases. Each has two layers.
Layer A walks you through the mechanism in everyday terms. It ends with a single check question — stop before reading the answer, try to formulate your own, then read on. Layer A stands on its own. If you read only Layer A, you'll come away with a full understanding of how and why a dog switches between its senses.
Layer B draws on the scientific literature and digs deeper into anatomy. It gives numbers from studies, names of pathways and nuclei, and indicates which elements have been measured directly in the dog and which are extrapolated from rodent, primate, or human models. Layer B also stands on its own. If you have the background, you can read just this layer.
My suggestion: read all of Layer A first. When you feel you understand the mechanism, come back and work through Layer B. But you don't have to stick to that — both layers can be read in any order.
Phase 1. A dog in full scent mode
[A] How it works
When a dog is working in full scent mode, it breathes differently from when it's just walking. It's a separate physiological act. The frequency of sniffing rises several-fold — from a resting rhythm, the kind you and I breathe at, to a fast, short, staccato one. You can hear it from a metre away, because it generates a turbulent stream of air in the dog's nose. I didn't expect to find elements I know well from micromechanics and diving... in Mrok's nose.
A dog's nose isn't an ordinary tube where air just comes in and goes out. The inhale goes one way — straight to the back of the nasal cavity, where there's an extensive area of olfactory epithelium waiting, the place where scent molecules meet receptors. The exhale, on the other hand, is directed downward and out to the sides, through the lateral slits of the nostrils. Why this separation? So that the exhaled air doesn't flush the scent out of the area that's still being analysed. And more interestingly: the exhaled stream, on its way out, lifts scent molecules from a few centimetres in front of the nose and draws them into the next inhale. A dog literally pulls in scent that, without this manoeuvre, would be beyond its reach.
The practical consequence is active sniffing, not passive breathing. In this mode, the dog isn't just inhaling — he's actively working his nose like a pump that supplies its own fuel for analysis.
Looking deeper into the dog's nose... beyond that epithelial zone, a network begins whose scale is hard to grasp intuitively. Where a human has a few million olfactory sensory cells, a dog has around two hundred million. Where in a human the brain has a small, distinctly modest area dedicated to smell, in a dog the same area is, in proportion to the whole brain, about thirty times larger. This isn't "a slightly better nose" — it's a different category of organ, occupying a place in the architecture of the dog's brain that a human can't even imagine.
But the key to this story isn't how well a dog smells. The key is what the olfactory network does with the information, where that information goes directly, and where it has to pass through a checkpoint.
Here we come to the first unusual fact, which organises the rest of this article. Smell, in all mammals, not just dogs, is the only sense that has direct access to the cerebral cortex, bypassing the first checkpoint every other sense has to pass through. Hearing, sight, touch, taste — all of it gets filtered, before reaching the cortex, through a structure in the centre of the brain that works like the on-duty controller at an air traffic tower. The controller decides which signals to let through at full strength and which to quiet down. Smell doesn't need the controller's permission. It goes through directly.
This is one reason why a dog who is intensely sniffing something is hard to call off. Olfactory information isn't regulated by the same mechanism that regulates the other senses. I'll come back to this in phase three — it's the foundation of everything that follows.
One important point before I close out Layer A of this phase. Active sniffing has a hidden cost. The fast, turbulent stream of air in the nose generates its own noise — audible to the dog itself. It's the same effect you get when you breathe quickly through your nose — sounds from outside start to disappear under the sound of your own breath. (I'd recommend trying it: start breathing very quickly through your nose while quietly listening to a podcast.) In a dog, the effect is far stronger, because the rhythm is faster and more forced. A dog who is sniffing intensely hears worse — not because his ear works differently, but because his audio input is noisy.
Stop before answering. If your dog doesn't react to your call during intense sniffing, what's the most likely reason — and is there more than one?
A moment for your own answer 😊
There are two, overlapping reasons. The first, immediate and physical: the dog's own breathing generates noise that lowers his ability to detect quieter sounds from outside. Your call competes with his sniffing — and often loses. The second, deeper reason: smell is the only sense that bypasses the primary control over access to the cortex, so when a dog is sniffing, his system can't just "turn smell down" to make room for the other senses. Hearing, in this moment, is regulated differently — it's operating from a worse starting position. I'll show the mechanism in detail in phase three. For now, it's enough to remember that "the dog isn't listening" and "the dog can't listen on equal terms in this mode" are two different things.
[B] What exactly is happening
Active sniffing in a dog is not breathing. It is a separate physiological act, mechanically and aerodynamically decoupled from respiration in a way that can be surprising.
When a dog is working in full scent mode, the frequency of its "breaths" rises from about 2–4 Hz at rest to 5–12 Hz, depending on the task and sampling phase. But what matters is not the breathing rate but the path the air takes through the dog's nose. Brent Craven's group (initially at the US Naval Academy, later the FDA) showed, on the basis of computed tomography, schlieren imaging (the streak method) and CFD simulations (computational fluid dynamics — the same way you check, for example, the aerodynamics of a car body), that the dog's nose has two separate paths for inhaled and exhaled air. The inhale goes through the upper part of the nostril straight to the olfactory zone at the back of the nasal cavity — where the olfactory epithelium (neuroepithelium, stretched over an area on the order of several tens of cm² in a dog) meets the odour molecules. The exhale is directed downward and to the sides through the lateral slits of the nostrils — so as not to flush the zone back out and not to disperse the scent the dog is in the middle of analysing.
It is a very interesting system. The exhaled stream generates a turbulent jet that, from the outside, lifts odour molecules from a distance of a few centimetres in front of the nose and delivers them toward the next inhale. The dog doesn't just sniff what is under its nose — it actively draws in scent that, without this manoeuvre, would be out of its reach. In experiments by Staymates et al. (Scientific Reports, 2016) with 3D-printed models of a dog's nose connected to sensitive chemical vapour detectors, active sniffing produced an 18-fold increase in detectability compared with a continuous inhale of the same volume. A sixteen-fold increase was obtained by attaching the dog's-nose principle to a commercial explosives detector (a technical note here: I found different values in several places, so instead of "18 times" it is safer to say that active sniffing multiplies the efficiency of delivering material for analysis many times over).
The olfactory epithelium in a dog contains, depending on the breed, from 125 million (in short-nosed breeds) to 300 million (in the Bloodhound) olfactory sensory neurons (ORN, olfactory receptor neurons), with an average of around 220 million. For comparison, in humans this is about 5–6 million. Each ORN has immobile olfactory cilia at its apical end, in whose membrane sit protein receptors sensitive to a specific type of odour molecule. In a dog there are about 800 different functional types of such receptors, in a human about 350. Each ORN expresses only one type of receptor. The ORN axons pass through the cribriform plate of the ethmoid bone (lamina cribrosa) and reach the olfactory bulb (bulbus olfactorius), which in a dog is a structure disproportionately larger than in us — proportionally about 30 times larger relative to the whole brain.
In the bulb, ORN axons form spherical structures called olfactory glomeruli. The key rule: all ORNs expressing the same type of receptor converge onto the same glomerulus (or onto a small group of glomeruli). This means a scent is laid out across the bulb as a unique, spatial pattern of glomerular activation — a kind of fingerprint in the bulb's topography. Each glomerulus contains about 1000 ORN axons synapsing with 20–50 mitral cells and tufted cells — second-order neurons that send the information onward. One mitral cell has one primary dendrite going to one glomerulus (the "one cell, one glomerulus" rule), but its axons fan out widely across the olfactory cortex. This produces a massive convergence of information: 1000 ORNs are funneled into one signal from a mitral cell, giving a 1000-fold amplification of the signal. Here I'm reminded of night vision, which I use for observing animals — both it and the olfactory bulb work on the same principle. A single weak signal is lost in the noise, but converging a thousand such signals at one output point pulls it above the detection threshold.
The axons of the mitral and tufted cells leave the bulb via the lateral olfactory tract (LOT). They reach several targets: the piriform cortex, the anterior olfactory nucleus, the olfactory tubercle, the entorhinal cortex, the peri-amygdaloid cortex, and directly the amygdala. And here's the catch: olfactory information does not have to pass through the thalamus before reaching the cortex. It is the only sense that enjoys this privilege. All other senses (hearing, sight, touch, taste) must first go through the thalamus, and in the thalamus sits the thalamic reticular nucleus (TRN), a layer of inhibitory neurons that functions as an attention filter. Smell goes directly, skipping the queue.
(More precisely: the thalamus does participate in later olfactory processing — the mediodorsal nucleus of the thalamus receives information from the piriform cortex and amygdala, and has reciprocal connections with the orbitofrontal cortex. But this is a second-order relay, not the primary gate. Smell has direct access to the olfactory cortex and the amygdala before the thalamus even finds out about it.)
And now the punchline — or more precisely the big WOW for me as an amateur, and I'm keeping my fingers crossed for the continuation of the research 😊:
In 2022, a team from Cornell — Erica Andrews, Raluca Pascalau, Alexandra Horowitz, Gillian Lawrence and Pip Johnson — published in the Journal of Neuroscience the first detailed map of white-matter pathways leaving the dog's olfactory bulb. It was performed ex vivo with DTI tractography on 23 brains and validated by Klingler anatomical dissection. The team found five large tracts: to the piriform lobe, to the entorhinal cortex, to the limbic system, to the corticospinal tract, and — the surprise — to the occipital lobe (i.e. the visual cortex). This last pathway has not yet been described in any other species. In an anatomical sense, the dog has direct wiring from smell to the area that in other mammals is dedicated exclusively to image processing. The functional meaning of this pathway is still an open research question. One of the explanations being considered covers situations like a blind dog effortlessly finding or catching thrown objects, but direct evidence is not yet available. For my own purposes I've adopted the phrase that "the dog sees with its nose."
Hearing, meanwhile, works differently. A sound wave enters the pinna, travels along the auditory canal, hits the eardrum, passes through the ossicles (malleus, incus, stapes), excites the hair cells of the cochlea, is encoded electrically, travels via the vestibulocochlear nerve (VIII) to the cochlear nuclei of the brainstem, from there to the superior olive, to the nuclei of the lateral lemniscus, to the inferior colliculi of the midbrain, to the medial geniculate nucleus (MGN) of the thalamus, and only from there to the primary auditory cortex (A1). Each of these stations is an opportunity for modulation — for amplifying the signal or damping it. And the key station from the point of view of attention is the MGN in the thalamus, because the TRN is draped exactly over it.
The TRN is a thin layer of GABAergic neurons (i.e. inhibitory, secreting γ-aminobutyric acid), laid like a cap over the thalamus. Francis Crick in 1984 called it the "gatekeeper" in a classic PNAS paper. When an animal is to direct its attention to the visual modality, the visual sector of the TRN lowers its inhibition over the lateral geniculate nucleus (LGN, visual), letting the visual signal flow more freely to the cortex — and at the same time the auditory sector of the TRN raises its inhibition over the MGN, damping hearing. The mechanism is measurable: in optogenetic fMRI in mice (Wimmer et al., Nature 2015; Halassa et al., Cell 2014) exactly this signal reversal was shown. The TRN is a mechanism for gating attention between modalities — with the one mentioned exception that smell bypasses this gating at the first level.
What does this mean in practice for a dog sniffing the ground?
Its system cannot "turn smell down" in the same way a mammal turns down one of the senses passing through the thalamus. Smell is regulated differently — at the level of the bulb (via periglomerular and granule cells, which provide local feedback inhibition) and via centrifugal (efferent) projections from the olfactory cortex back to the bulb (a less understood but significant mechanism). Modulation of hearing during active sniffing, on the other hand, happens via the standard TRN mechanism: hearing is gated, the detection threshold for sounds rises. It isn't switched off. It simply gets a lower priority.
Add to this the aerodynamics of sniffing itself. Turbulent exhalation bursts at 5–12 Hz are a continuous, low-level noise in the dog's airways, audible to the dog itself. The signal-to-noise ratio for receiving external sounds drops. The sum of these two mechanisms — gating at the TRN level plus the dog's own aerodynamic noise — produces an effect that from the outside looks like "the dog can't hear" and in reality is "the dog has severely degraded auditory input in two independent places."
Phase 2. The moment of detection
[A] How it works
In the stream of molecules the dog is drawing into its olfactory zone, something appears that wasn't there before. A fresh deer trail. The scent of a familiar dog. A subtle note of fermentation. The signature of a person outside the standard repertoire. It doesn't have to be a single compound — a dog's brain doesn't analyse scents element by element, like a chemist breaking a sample down into its components. It analyses them as a whole, as a pattern, as a configuration. "This forest at this time of year smells like this" — the dog has had that pattern in his head since the start of the walk. And what comes in with each new inhale is compared against that pattern.
When everything matches, the system is calm. The dog walks on, sniffs on, nothing happens. When something doesn't match, when the molecular stream carries something the pattern didn't predict, an alarm goes off. The alarm here isn't a metaphor. It's a concrete reaction of the amygdala, a structure deep in the brain whose job is to answer the question "is what just came in important?" In a dog, the answer comes fast. Scent information reaches the amygdala by a short path, with a minimal number of transfers — much shorter than for any other sense. This is another structural fact that organises a dog's whole life: smell reaches the centre of emotional evaluation before the dog can "think" in any sense of the word.
When the amygdala signals "important," it triggers a cascade. The signal goes to a small structure in the brainstem — the locus coeruleus. The structure is small (a few tens of thousands of neurons), but its axons branch so widely that it can reach vast areas of the cortex at once. It isn't a relay station. It's a global mode switch. When the locus coeruleus fires a short, intense burst, the whole cortex changes how it operates: strong signals get stronger, weak ones get suppressed, the system becomes more selective, more ready to interrupt the current task and do something else.
What we colloquially call the "orienting reflex" — the dog freezes, turns his head toward the stimulus, stops doing what he was doing — is the body's reaction to a mismatch between what the brain predicted and what came in through the senses. The Soviet neurophysiologist Yevgeny Sokolov described this mechanism systematically in his 1963 monograph and showed its key feature: the reflex fires only when the stimulus diverges from prediction. Not on every stimulus. Only on the one that doesn't fit the model.
From this, a simple but important conclusion follows. The dog's brain is a prediction machine. Throughout the whole walk it produces a forecast: "what should I smell, see, hear in this place next?" Sensory inputs are compared against that forecast. The alarm fires only when something doesn't match.
Stop before answering. Why doesn't a dog orient to every sound and every smell around him, when in a forest there are thousands of them at once?
A moment for your own answer 😊
Because his brain doesn't react to stimuli as such — it reacts to the mismatch between a stimulus and the prediction. The thousands of sounds and smells surrounding the dog in the forest are part of the model he holds in his head. They fit. They don't generate an alarm. The alarm fires only when something doesn't fit the model — a new scent signature, a sound in the wrong place, silence where there should be a rustle. The system is built to ignore what's predictable and to react to what stands out. If it were otherwise, the dog would be permanently overloaded.
[B] What exactly is happening
The dog's brain analyses mixtures of scents configurally, as a whole, rather than elementally, decomposing them into components. Research by Gregory Berns's team at Emory University (Prichard et al., Chemical Senses, 2020), carried out on 18 conscious dogs in an fMRI scanner, showed this directly. For comparison, most research on human smell suggests rather elemental decomposition. The dog's olfactory system is trained in recognising patterns, not components. At the neural level this is the work of the piriform cortex: a single pyramidal cell of the APC (anterior piriform cortex) receives convergent input from at least four different glomeruli of the bulb, and its activation requires the synchronous arrival of this input. This is the principle of population coding: a scent is not a single signal but a pattern of activity across a distributed network.
Olfactory information in a dog reaches the amygdala in two ways — directly, across a short synaptic distance, to the cortical nucleus, and via the piriform cortex to the basolateral nucleus (BLA). In the BLA the brain links a stimulus (e.g. a scent) with its emotional meaning (reward, threat, indifference). The Berns team's experiment showed that already after the first exposure the dog's amygdala distinguished a scent associated with reward from a neutral scent — and did so more strongly than the cortical area responsible for cognitive analysis. In other words: the emotional evaluation of a scent in a dog is faster than conscious recognition. The body starts reacting before the dog has time to "think." With all the caution with which the word "think" has to be used in the case of a dog.
When the amygdala signals relevance, a cascade follows. The signal goes to the locus coeruleus (LC) — a small, densely packed nucleus in the brainstem, containing only a few tens of thousands of neurons. But these neurons are noradrenergic (they produce noradrenaline, NA), and their axons branch out so widely that a single LC neuron can innervate vast stretches of the cortex. This is not a relay station. It is a global modulator of the brain's state.
According to the model of Gary Aston-Jones and Jonathan Cohen (Annual Review of Neuroscience, 2005, one of the most frequently cited papers in cognitive science of the last quarter-century), the LC operates in two modes: tonic and phasic. The tonic mode is a steady, low-intensity firing of the neurons — it keeps arousal at a general level appropriate to the current task. The phasic mode is a short, strong burst that appears when the brain encounters something relevant. That burst has a concrete, measurable effect: it interrupts the current cognitive task and forces the system to regroup. Noradrenaline acts on α1, α2, β1, β2 receptors scattered throughout the cortex and brainstem — the effect is a change in the gain of cortical neurons: strong signals become even stronger, weak signals get suppressed. Mathematically this corresponds to "sharpening" the system's selectivity.
This pattern was described in 1963 by the Soviet neurophysiologist Yevgeny Sokolov in the monograph Perception and the Conditioned Reflex (Pergamon Press, Oxford). He called it the orienting reflex and showed that it consists of a characteristic set of components: halting the current behaviour, turning the head or body toward the stimulus, a change in heart rhythm (first a brief slowing, then an acceleration), pupil dilation, an increase in skin conductance, activation of muscles ready for action. Sokolov also showed that the reflex is selective. It does not react to every stimulus; it reacts only when the stimulus does not fit the current model of the world the dog has internalised. In other words, it is triggered by the mismatch between what the system expects and what it actually gets.
At every moment the cortex generates a model of what is about to arrive, and sensory inputs are compared against that model. When something matches, no alarm. When something deviates from the model — alarm, orienting, noradrenaline, mode switch. This is a foundation of contemporary cognitive science, known as predictive coding.
On the footage from Mrok's camera, I see this as the moment when the dog stops breathing.
Phase 3. Three seconds of switching
[A] How it works
At a certain moment in the recording, I could hear Mrok's breathing go quiet. It doesn't stop completely — he's still breathing, but the sniffing rhythm fades. The inhale lengthens, the exhale grows shallower, the tempo drops. This happens because the signal-to-noise relationship works in both directions. If you want to hear better, you have to quiet your own noise. It's the same reflex you have when you strain to hear footsteps in a dark corridor and hold your breath. Not because someone taught you to, but because it's built into the brainstem of every mammal. Both the amygdala and the cortex have a direct line to the respiratory centres in the medulla oblongata, and can briefly switch them off. It's an evolutionarily cheap function with a big payoff — the moment of respiratory silence is when hearing works at full capacity.
The dog brings his ears into play. A dog's pinna isn't dead cartilage. It's an active, directional microphone, steered by a dozen or so small muscles. The two pinnae move independently of each other, letting a dog triangulate the source of a sound with a precision a human can only dream of — our ears are stuck to our skull. A dog doesn't have to turn his head to change the direction of reception. When he rotates one ear, he sets the axis of maximum sensitivity of that pinna in a specific direction. The other ear can cover a different area at the same time. It's the equivalent of a two-antenna radar with independent steering of both antennas.
The effect of the ears' work is most visible in dogs with upright ears like German Shepherds, Huskies, or ACDs. In dogs with drop ears the mechanism works the same way at the muscle level; it's just less visible from the outside. The muscles are working, steering, even if from our perspective the pinna seems not to move.
The third thing, invisible in the recording: the filter governing which senses are used switches over. The same mechanism that in phase one kept hearing on low priority now gets the opposite signal. A noradrenaline burst from the locus coeruleus modulates the filter's operation. The auditory sector reduces its inhibition — the auditory signal flows to the cortex more widely and more strongly than three seconds earlier. The detection threshold for sounds drops dramatically. The dog doesn't hear "more sharply" in the physical sense of ear sensitivity — the inner ear works exactly as it did a minute ago. What changes is the selectivity of the cortex. Sounds that were background a second ago are now being analysed. The cortex starts applying specific filters tuned to the hypothesis smell has put forward. "The source is over there. What features of the sound will confirm or rule out what scent suggested?"
Then the rest of the body synchronises with the work of attention. The head lifts or turns slightly. The tail freezes. The neck and torso muscles tighten subtly — the body shifts into a state of readiness, but not yet action. The pupils dilate (this is one of the most reliable indicators of locus coeruleus activation, measurable in humans too). The pulse briefly rises. The dog isn't thinking about any of this. His body reconfigures without conscious involvement, as a single integrated system.
At the same time, smell doesn't switch off. The dog doesn't choose "either sniffing or listening" — he integrates. Smell is still working in the background, receptors still pick up molecules, the olfactory bulb still encodes them. But cortical priority now belongs to hearing. Data from both senses flow into integrating areas, including one very particular place — the olfactory tubercle, a structure that, in addition to scent information, also receives auditory input. That's where the dog's brain answers the question of whether to confirm the result of the olfactory analysis. "Scent suggests a deer. Does the sound confirm it?" The integration happens on a timescale measured in fractions of a second.
An important caveat for those who like precise data: we don't have precise switching-time measurements for the dog. The figures on the order of hundreds of milliseconds known from the literature come from studies in humans and primates. In the dog, measuring quantities like these in an fMRI scanner is technically difficult: a dog won't hold still for an hour, and even if he did, the structures of interest are at the edge of the equipment's resolution. What we know for sure is that the mechanisms are homologous — all mammals use the same set of structures, differing mainly in proportions and connection density. So "order of magnitude: hundreds of milliseconds" is a good answer.
Stop before answering. During the switching phase, is the dog "distracted"?
A moment for your own answer 😊
No. He's focused — just on a different modality. From the outside it looks like someone who was just doing something intensely and suddenly stopped — and it's easy to read that as "a lapse in attention." But underneath, exactly the opposite is happening: the system is carrying out a coordinated operation — switching priority between senses, silencing its own noise source, reconfiguring the cortical filters, redirecting the work of the pinnae, changing the rhythm of heart and muscles. A dog that can do this fluently, and can return to the original task afterwards, has a well-functioning attention system. Distraction is something else: chaotic switching without a return to base, without a reception phase, without a resolution. The difference is subtle from the outside, but cognitively it's fundamental.
[B] What exactly is happening
The first measurable change is the quieting of the breath. The sniffing pattern fades, the frequency drops. The regulation happens directly through modulation of the respiratory generator in the medulla oblongata, mainly the preBötzinger complex and the surrounding groups of neurons. Higher tiers of the nervous system have direct input to this generator: the amygdala and the prefrontal cortex can silence breathing for a short time, because it is an evolutionarily valuable function. The circuit is conserved across all predatory mammals.
The second change is the work of the pinnae. The dog's pinna is steered by groups of muscles — veterinary classifications give "more than 12" — organised into dorsal, rostral, ventral and caudal components. All innervated by the facial nerve (VII). Each pinna works independently — it can be set at a different angle from the other, which gives the dog a binaural localisation of direction that is unavailable to humans. In breeds with drop ears the mechanism works identically at the muscle level; only the visual effect is weaker.
The third change, invisible from the outside, happens at the TRN level. The noradrenaline burst from the LC modulates TRN activity mainly via α receptors (α1 depolarizing, α2 hyperpolarizing on TRN neurons). The auditory sector of the TRN lowers its GABAergic inhibition over the MGN, which allows a wider flow of signal to the primary auditory cortex A1. The detection threshold for sounds drops. The resolution of selective auditory attention rises — the cortex begins to apply narrowband acoustic filters, tuned by top-down feedback from the dorsolateral prefrontal cortex (dlPFC) to fit the current hypothesis about the source. This is the classic attention-gating mechanism, described in humans and primates, extrapolated to the dog by homology of structures.
The fourth change is the full activation of the somatic component of the orienting reflex. The nucleus of the solitary tract (nucleus tractus solitarii, NTS) in the brainstem modulates heart rhythm — through the vagus nerve (X) a brief slowing occurs, then through the thoracic sympathetic trunk an acceleration. Pupil dilation is the effect of noradrenergic activation of the pons — it is such a reliable indicator of LC activity that in cognitive research on humans pupillometry is used as a surrogate measure of the LC. In the dog the effect is identical, though not always easy to observe because of iris pigment.
The fifth change is multisensory integration. Data from the two modalities converge in several areas: the superior temporal sulcus, the temporoparietal cortex, the dorsolateral prefrontal cortex, and — particularly interesting for the dog — the olfactory tubercle. The olfactory tubercle, besides projections from the bulb, receives direct auditory inputs. Wesson and Wilson (Journal of Neuroscience, 2010) in a paper titled "Smelling sounds" showed that in rodents single neurons of the olfactory tubercle encode both olfactory and acoustic features simultaneously — the first such direct evidence of chemical-acoustic integration at an early stage of processing. In the dog, whose olfactory tubercle is proportionally much larger, the mechanism is probably analogous, although direct measurements are not yet available.
A technical methodological note: the times of 200 ms for TRN opening, 500 ms for switching A1 selectivity, about a second for full multisensory integration — these are figures from studies on humans and primates, with isolated extrapolations to rodents. In the dog, direct in vivo electrophysiological measurements of the TRN have not been carried out. Measurements of the auditory threshold during active sniffing in the dog are also not available as a psychoacoustic experiment. Confirmation rests solely on observation. The mechanisms are homologous across species, and in the dog they are confirmed anatomically (DTI), neuroimaging-wise (the Berns team's fMRI) and behaviourally. Direct timing measurements in the dog do not exist.
Phase 4. Decision
[A] How it works
When the cortex receives data from the newly prioritised modality, it compares it against what it already holds in its internal library — patterns of scents, sounds, sights, situations the dog has met in its life. An adult dog's library is enormous. Every walk adds entries to it. Every interaction with another dog, every encounter with a cyclist, every deer passed on the horizon — it's all recorded in there. And now, in the third second of the freeze, the dog is matching what it has just detected against that library.
There are three outcomes.
Outcome one: nothing. The signal matches the "known, unimportant" pattern. A woodpecker hammering a tree trunk fifty metres away. Wind in the branches. A hare that's already run past. In this case the dog returns to sniffing. Exactly the same rhythm, the same place, as if nothing had happened. Noradrenergic arousal fades, the thalamic filter returns to its olfactory settings, the auditory threshold rises again, sniffing resumes at the rhythm of three seconds ago. The only difference: the library has just updated with a new entry — "I've now recognised this specific acoustic signature in this specific place."
Outcome two: opportunity. A fresh game trail with a characteristic sound. A familiar dog on a neighbouring path. In this case the dog changes his behaviour: the modalities become fully integrated, smell and hearing working together toward a single goal, with vision joining in. Tracking begins, then the approach, then the chase.
Outcome three: threat. An atypical signature that can't be categorised as safe. A boar at a distance from which there's no sensible retreat. A person behaving unsettlingly. A full stress response kicks in — the hormonal axis, cortisol release, the body prepared for fight, flight, or freeze. This is no longer the orienting reflex but a qualitatively different state, with a different hormonal profile and a different duration (minutes, not seconds).
Over an hour-long walk with Mrok, the orienting reflex fired — counting only what was caught on camera — at least 30 times. Over half of those led to the first outcome. Woodpeckers, squirrels, wind in the leaves. A quarter — a deer or some other animal that neither he nor I saw, but which had left a trail strong enough to trigger an alarm, then turned out to be out of reach. The rest are things classified in his internal library as "known." He went back to sniffing every time.
And this is the key thing for interpreting behaviour. The first outcome — the apparently uninteresting "nothing happened" — is actually the most impressive. Switching the attention system, gathering data from another modality, matching against the library, finding no alarm pattern that fits, cancelling the alarm, returning to the original task in a few seconds without stalling and without dragging out the tension — that's the mark of a very well-regulated brain. A dog that gets stuck in phase three, starts barking, and can't return to sniffing has a regulation problem. A dog that moves through phases one to four fluently, hundreds of times a day, has a functioning attention system.
And here begins the practical part — the part that gives this text a point beyond curiosity.
A dog that doesn't respond to your calls during intense sniffing isn't "stubborn" or "disobedient." He's in a different cortical mode, with a raised auditory threshold — exactly the one I described in phase three, only with the sign reversed. When smell has priority, hearing gets a higher threshold. If you want him to react, you have two sensible options. First: generate a signal strong enough to break through that threshold (a louder call, a whistle, a clap). Second (usually the better choice): wait. Most "disobedient" dogs in the field are simply dogs who happen to be in a work phase.
A dog that can move between phases fluently has a valuable competence: flexibly switching attention back to base. It's one of the things good working-dog trainers evaluate first when looking at a young dog. Not "does the dog respond to a command?" but "can the dog return from high arousal to lower arousal?" These are not the same thing.
Very important: when we say a dog "lives by its nose," that's true, but it's an incomplete truth. A dog lives by smell as his dominant sense. But his attention system isn't jammed on smell. At any moment he can set that priority aside and point — now hearing, now vision, now back to smell. This ability to switch priorities fluidly between modalities, and to return elastically, is an executive function the dog has acquired through tens of thousands of years of selection for cooperation with humans. Not every species has it. In the dog, it's clearly better developed than in most of the animals that have shared the evolutionary road with us.
Stop before answering. Dog X doesn't react to the recall command in the field, though at home he responds flawlessly. Dog Y always reacts in the field, regardless of what he was doing a moment before. Which of them has the better attention system?
A moment for your own answer 😊
Neither description is enough to decide. This is a trick question 😊. Dog X may have an excellent attention system working in olfactory mode with a raised auditory threshold. Without a stronger signal, there's no way to pull him out of that mode. That's not a flaw, just a variant of operation. Dog Y may have an attention system with weakened focus, too susceptible to external signals, incapable of deep work in any modality. That's not an advantage either — just a different variant. The key question isn't "does the dog react," but "after high arousal, can he return to base?" Dog X, who finishes sniffing, lifts his head and comes back to you normally, is well regulated. Dog Y, who reacts immediately but can't sustain work in scent for longer than thirty seconds, is poorly regulated. Training a response to a command is only one of many axes on which an attention system is assessed — and… not always the most important one.
[B] What exactly is happening
Comparison of the incoming signal with patterns in long-term memory is a process distributed across several structures. The hippocampus (with inputs from the entorhinal cortex) participates in access to episodic memory — "does what I'm detecting now resemble something I've seen before?". The prefrontal cortex (proportionally smaller in the dog than in primates, but functionally present) integrates the data and compares it with the current behavioural context. The amygdala adds the emotional evaluation. The basal ganglia modulate the choice of response. The decision is the result of the convergence of all these signals.
Three possible outputs correspond to three different neurochemical profiles:
Output irrelevant → drop of LC activity to a tonic level, fading of the noradrenergic burst, unblocking of the TRN for the primary modality (smell), return to the previous state. The signal is added to the habituation patterns — next time the same stimulus will generate a weaker LC response (the classical habituation already described by Sokolov).
Output opportunity → maintenance of tonic LC arousal at an elevated level, engagement of the dopaminergic system (the mesolimbic pathway, the nucleus accumbens, the ventral tegmental area VTA) — which encodes the expected reward. Multisensory integration is maintained. Motor activity is switched into goal mode — through the pyramidal system and modulation of the basal ganglia.
Output threat → activation of the HPA axis (hypothalamus → pituitary → adrenal glands), release of corticotropin, cortisol, catecholamines. Full engagement of the sympathetic system. The central nucleus of the amygdala (CeA) projects to the periaqueductal grey (PAG) and to the brainstem, simultaneously triggering freezing, or the components of fight, or flight. This is a state qualitatively different from the orienting reflex — different receptors, different duration (minutes, not seconds), different behavioural consequences.
Regulating the return from high arousal to the baseline state — the ability to return to base — is one of the key indicators of the health of a dog's emotional system. Mechanistically it is governed above all by the inhibitory projections from the ventromedial prefrontal cortex (vmPFC) to the amygdala and by the GABAergic system at many levels. A dog with impaired regulation in this circuit — as a result of developmental stress, insufficient socialisation, chronic arousal during the adolescent period — has difficulty extinguishing the alarm response even after the stimulus has passed. From a behavioural perspective this looks like "hyperactivity" or "lack of concentration" — but the neural substrate lies in a specific, measurable circuit.
The same applies to modality switching. The ability to fluently switch attentional priority — described by Michael Posner and others in the context of the human attention system as one of three components (alerting, orienting, executive control) — is a function of the cortico-subcortical attention network comprising the prefrontal cortex, the parietal sulcus, the TRN and the LC. In the dog this network is present, functionally less studied than in primates, but anatomically homologous. Assessment of its efficiency in a particular dog, on the basis of behavioural observation, is possible and practical, but it requires observation in many contexts, not just in a single training situation.
What to take away
Three seconds during which a dog stops breathing are a window onto one of the oldest mechanisms of the mammalian brain — and onto how that mechanism works in a dog with a precision that's hard to appreciate without camera footage. A dog's attention system isn't "worse" or "simpler" than a human's — it's built around a different dominant modality, with different neural pathways, with smell in a privileged position. Interpreting a dog's behaviour without understanding this architecture leads to errors that, in training practice, can be costly: "disobedience" may be work, "distraction" may be focus on a different modality, "stubbornness" may be an anatomically enforced auditory threshold. The most important conclusion comes from repeated observation, not a single scene: a healthy attention system is recognised by its return to base, not by its reaction to the stimulus.
And the collar camera, bought with the idea of recording Mrok's interactions with other dogs, turned out to be a cognitive tool at a far higher level than I'd expected when buying it.
By Wojtek A. Filip (Paws-to-Peaks). English translation by Karolina Jacobss-Fea, everyday carer and trainer of Dougal, a Bearded Collie × Border Collie cross. References
Key works underpinning the claims in this text:
Andrews, E.F., Pascalau, R., Horowitz, A., Lawrence, G.M., Johnson, P.J. (2022). Extensive connections of the canine olfactory pathway revealed by tractography and dissection. Journal of Neuroscience, 42(33), 6392–6407. DOI: 10.1523/JNEUROSCI.2355-21.2022
Aston-Jones, G., Cohen, J.D. (2005). An integrative theory of locus coeruleus-norepinephrine function: adaptive gain and optimal performance. Annual Review of Neuroscience, 28, 403–450.
Craven, B.A., Paterson, E.G., Settles, G.S. (2010). The fluid dynamics of canine olfaction: unique nasal airflow patterns as an explanation of macrosmia. Journal of the Royal Society Interface, 7(47), 933–943. DOI: 10.1098/rsif.2009.0490
Crick, F. (1984). Function of the thalamic reticular complex: the searchlight hypothesis. Proceedings of the National Academy of Sciences, 81(14), 4586–4590.
Ferreira-Cardoso, H., Silva, A.C., Pereira das Neves, J., Silva, I., Faustino-Rocha, A.I., Colaço, B., Oliveira, P.A., Gama, A. (2023). The olfactory bulb in companion animals: anatomy, physiology, and clinical importance. Brain Sciences, 13(5), 713.
Halassa, M.M., Chen, Z., Wimmer, R.D., Brunetti, P.M., Zhao, S., Zikopoulos, B., Wang, F., Brown, E.N., Wilson, M.A. (2014). State-dependent architecture of thalamic reticular subnetworks. Cell, 158(4), 808–821.
Kokocińska-Kusiak, A., Woszczyło, M., Zybala, M., Maciocha, J., Barłowska, K., Dzięcioł, M. (2021). Canine olfaction: physiology, behavior, and possibilities for practical applications. Animals, 11(8), 2463. DOI: 10.3390/ani11082463
Posner, M.I., Petersen, S.E. (1990). The attention system of the human brain. Annual Review of Neuroscience, 13, 25–42.
Prichard, A., Chhibber, R., Athanassiades, K., Spivak, M., Berns, G.S. (2020). Decoding odor mixtures in the dog brain: an awake fMRI study. Chemical Senses, 45(9), 833–844. DOI: 10.1093/chemse/bjaa068
Sokolov, E.N. (1963). Perception and the Conditioned Reflex. Pergamon Press, Oxford.
Spence, C., Driver, J. (Eds.) (2004). Crossmodal Space and Crossmodal Attention. Oxford University Press.
Staymates, M.E., MacCrehan, W.A., Staymates, J.L., Kunz, R.R., Mendum, T., Ong, T.-H., Geurtsen, G., Gillen, G.J., Craven, B.A. (2016). Biomimetic sniffing improves the detection performance of a 3D printed nose of a dog and a commercial trace vapor detector. Scientific Reports, 6, 36876.
Wesson, D.W., Wilson, D.A. (2010). Smelling sounds: olfactory-auditory sensory convergence in the olfactory tubercle. Journal of Neuroscience, 30(8), 3013–3021.
Wimmer, R.D., Schmitt, L.I., Davidson, T.J., Nakajima, M., Deisseroth, K., Halassa, M.M. (2015). Thalamic control of sensory selection in divided attention. Nature, 526(7575), 705–709.
Review and reference sources: Evans, H.E., de Lahunta, A., Miller's Anatomy of the Dog (standard veterinary reference); MSD Veterinary Manual, anatomy of the dog ear; Frontiers in Veterinary Science (2018, 2023), canine olfaction, aerodynamics of active sniffing; Bear, M.F., Connors, B.W., Paradiso, M.A., Neuroscience: Exploring the Brain, 4th ed. (foundations of sensory-system neuroanatomy).
Methodological note: some of the processes described here (direct measurements of auditory threshold during active sniffing in the dog, precise measurements of modality-switching latency in the dog, in vivo TRN activity in the dog) are extrapolated from rodent, primate, and human models. For the dog we have solid anatomical data (DTI, Klingler dissection), neuroimaging data (fMRI in awake dogs, Berns' team and others), and behavioral data. Direct electrophysiological experiments linking active sniffing with sound-detection threshold in the dog have not yet been published. The mechanism described is nonetheless consistent across species and confirmed observationally in the dog, including via the collar camera 😊, which, as it turns out, is itself a cognitive tool far more motivating to investigate than I had expected when I decided to start using such a camera. © Wojtek A. Filip / Paws-to-Peaks. You may quote and share with attribution and a link to the source. No reproduction of the full text without permission.
Comments