Cyclic Sighing: The Breathing Technique Your Body Already Runs

The double inhale you already know

You've done this before. Not as a technique - as a reflex. Think of the last time you cried hard, the ugly kind, until your chest hurt. When it was over, your body did something automatic: it stuttered. A big inhale, then a second one stacked on top, then a long exhale that seemed to go on forever. And then - a kind of settling. The storm had passed.

That stacked breath during and after crying is the physiological sigh. You didn't learn it, and you've run it from the very first days of life. A newborn sighs roughly 9 - 12 times an hour, and those early sighs do real work: the sigh-like breaths in the first hours are part of how the lungs finish inflating after birth. ^[1]11Jost 2015 You couldn't take a proper first breath without the pattern. From then on your body runs it automatically, for the rest of your life, without ever asking your permission. ^[2]22Severs & Vlemincx 2022

Here is the strange part. Every culture that ever noticed the sigh noticed only the way out - the long, sad exhale. The Japanese have a proverb that if you sigh, your happiness escapes. Baroque composers wrote a falling figure called the Seufzer, the musical sigh. Shakespeare's lovesick man sighs "like furnace." But the functional half of the sigh, the half that does the work, is the part nobody saw: the inhale, taken twice. We named the breath after its sound and missed what it was for.

What physiologists spent a century working out is that this involuntary reflex can be recruited on purpose. The surprise isn't that breathing helps - it's that your body already runs this exact pattern, on its own, as routine maintenance. A survival program baked into your brainstem, and you can take the wheel. Run it deliberately, five minutes a day, and it quiets your body's idle and appears to lift your mood - measurably - across an entire month. ^[3]33Balban 2023

Born in a WWI ward, not ancient

Here's what cyclic sighing is not: a rediscovered ancient practice. Search the old traditions and the deliberate double inhale isn't there - not in classical yoga, not as any named technique a tradition handed down for Western science to catch up with. The technique is entirely modern, built from a Western physiology lineage that begins in a World War I ward.

In 1919, J.S. Haldane and colleagues were studying British soldiers who had survived gas attacks. The survivors presented with a strange breathlessness - nothing organic, but their lungs felt tight. Lying flat made it worse. Haldane noticed that when they breathed shallowly for long enough, they began producing periodic large breaths automatically. He proposed those large breaths were re-expanding tiny sacs in the lungs - the alveoli - that had slowly collapsed under shallow breathing. ^[4]44Haldane 1919

That observation did not sit idle. By 1938, clinicians at the Mayo Clinic had characterized excessive sighing as its own syndrome - "sighing dyspnea" - noting it had no organic cause and was routinely mistaken for asthma or a cardiac complaint. ^[5]55Maytum 1938 Huberman has pointed to this same 1930s period as when scientists coined the term "physiological sigh," though no single paper or coiner for that exact phrase can be identified in the record. What is clear is that by mid-century the sigh had moved from curiosity to clinical syndrome. The next shift, in the 1990s, changed its shape entirely - from clinical complaint to neuroscience question. In 1991, a team led by Jack Feldman at UCLA identified the preBötzinger complex - a brainstem cluster that generates the respiratory rhythm, and the region later shown to house the dedicated sigh circuit. That paper ran in Science. ^[6]66Smith 1991 Then in 2016, Feldman's lab published in Nature: roughly 200 neurons, signaling through two peptides, turn out to trigger every sigh you've ever taken - a dedicated circuit, separate from ordinary breathing. (How precisely that circuit can be switched on and off is the subject of a later section.) ^[7]77Li 2016

By 2023, Feldman's co-author Mark Krasnow and Stanford psychiatrist David Spiegel had teamed up with Andrew Huberman to run the first randomized controlled trial of deliberate cyclic sighing. The result appeared in Cell Reports Medicine. ^[3]33Balban 2023

There's a claim in wellness spaces that cyclic sighing is the modern face of something ancient - specifically viloma pranayama, the interrupted-inhale yoga technique. The resemblance is real: both practices segment the inhalation. But the genealogy is backwards. Viloma first appears in B.K.S. Iyengar's Light on Pranayama in 1981; Iyengar was writing it down fresh, not translating a classical text. ^[8]88Iyengar 1981 And the science lineage has no yoga input anywhere in it - it runs from a WWI ward through two landmark papers in Science and Nature to a 2023 smartphone-delivered RCT. Convergent structure, not shared origin.

A century of physiology finally put the deliberate sigh to a controlled test in 2023. What's left is for other labs to repeat it, and for the last of the mechanism to be mapped - the work that turns a promising result into a settled one.

What cyclic sighing is for: reliable on the body, contested on mood

The claim you'll see most often, from Andrew Huberman's YouTube channel to wellness headlines, is that cyclic sighing is the fastest way to calm down in real time. ^[21]2121Huberman newsletter The evidence is more specific than that - and the specifics are interesting.

The body effect is reliable. Participants who practiced cyclic sighing for five minutes daily over 28 days breathed more slowly at rest - not just during the session, but throughout the day. Resting respiratory rate declined significantly more in the cyclic sighing arm than in the mindfulness control. ^[3]33Balban 2023 That's a physiological shift, not a psychological one.

The mood effect is real but shakier. In the same trial, the sighers came away a little happier than the meditators by the end of the month - 1.89 points of lift against 1.22 on the standard positive-mood scale. The catch is that only the sighers' gain was firm enough to trust; the meditators' could have been chance. ^[3]33Balban 2023; PANAS positive affect, sighing p = 0.025 vs mindfulness p = 0.06 On anxiety, the two finished in a dead heat - 3.95 points of relief for meditation, 3.85 for sighing, a gap you could lose in the noise. So when a headline says cyclic sighing "beats meditation," it's true for mood and wrong for anxiety. ^[3]33Balban 2023; STAI state-anxiety mean reduction −3.95 vs −3.85

The comparison that's missing - and this matters - is a placebo arm. In the largest breathing RCT ever run, 400 participants doing coherent breathing showed no significant benefit over a matched paced-breathing placebo. ^[9]99Fincham 2023 Cyclic sighing has never faced a sham condition. Its "best evidence" designation is real, but it means "best among studies that used an active control instead of a sham" - a narrower claim than most coverage implies.

Source: Balban et al. 2023, N = 108. ^[3]33Balban 2023 Between-arm differences are directional, not all statistically powered - study was not powered to prove superiority of any arm over another.

A case that fits less well: panic disorder.

If you've been diagnosed with panic disorder, or if ordinary breathing exercises tend to make you more anxious rather than less, this section is for you.

Panic disorder involves a specific breathing pattern: patients sigh more frequently than controls, run chronically low on CO₂, and - critically - recover CO₂ more slowly after each sigh. ^[12]1212Abelson 2001 Adding more deliberate sighs to a system that's already over-sighing and struggling to restore its CO₂ balance is moving the wrong direction. The mechanism papers that make cyclic sighing credible (Li et al., Nature 2016; Smith et al., Science 1991) are the same papers that explain why this specific pattern is a poor fit for this specific physiology.

Does it really work?

What held up is the body-calm story. People who practiced breathed more slowly at rest, and the slowing tracked with how much their mood improved - a real link, but a small one: the breathing change explains less than 6% of the mood swing. Signal, not magic. ^[3]33Balban 2023; resting-rate reduction p < 0.05; rate-mood correlation r = −0.24, p < 0.05 Mood itself rose more for the sighers than for any other breathing arm - the only group whose gain was statistically solid against meditation. ^[3]33Balban 2023; +1.89 ± 3.76 PANAS points, p = 0.025 And the body-calm pattern shows up outside Stanford too: an independent 2025 pain pilot ran cyclic sighing in an orthopedic clinic waiting room and found it cut pain intensity and unpleasantness - but, tellingly, left anxiety and depression untouched. ^[13]1313Hanley 2025; pilot RCT, n = 81, randomized 1:1 That null on mood, in a study designed to catch it, is exactly what the body-calm-over-mood-lift hierarchy predicts.

What didn't hold up is most of the rest. Three of the most-repeated claims simply didn't show up in the data: the trial moved neither heart rate, nor heart-rate variability, nor a single sleep measure, in any group. ^[3]33Balban 2023; HR, HRV, sleep all null across arms Anxiety relief was a tie, not a win - meditation was marginally ahead. And focus never got measured at all: the 2023 trial didn't look, and a 2026 observational study found sighs reset breathing variability during attention tasks but moved neither reaction time nor focus. ^[16]1616Observational 2026 "Cyclic sighing for focus" is hypothesis, not finding. For techniques with stronger attentional evidence, see the Focus hub.

That single-study foundation is why the honest grade is "moderate," not "strong" - what the field still needs is spelled out in For professionals, below.

How it works, and how to do it

The governing analogy for cyclic sighing is the relief signature. When danger passes - not during stress, but at the moment it ends - your body produces a surge of sighs. In humans, sighs cluster at relief rather than at peak stress. ^[10]1010Vlemincx 2009 In rats, the effect is striking: sigh rate ran about twenty times higher at relief than at quiet rest between trials - rat data, not a human figure. ^[11]1111Soltysik & Jelen 2005 (rats) The deliberate cyclic sigh may work partly by mimicking that signature: the brainstem reads the pattern as the "all clear" signal.

Under it are three layers:

Layer one: the invisible inhale. Alveoli - the tiny air sacs where oxygen crosses into blood - slowly deflate during ordinary shallow breathing. A single normal breath can't pop them back open; it doesn't generate enough pressure. The double inhale does. The first breath fills your lungs to about 70% capacity. The second inhale stacks on top, pushing lungs to full. That pressure difference reinflates the collapsed alveoli and restores gas exchange surface. ^[7,2]727Li 20162Severs & Vlemincx 2022 This is why Haldane observed it in WWI soldiers; it's why modern ICU ventilators still deliver periodic "sigh breaths" - without them, lung compliance deteriorates. ^[19,17]191719Feldman/Huberman17Mauermann 2014

Layer two: the exhale as brake pedal. When you breathe out slowly, your heart rate drops - a momentary deceleration you can feel as a small wave of quiet. The mechanism is respiratory sinus arrhythmia (RSA): as the chest expands on the exhale, intrathoracic pressure rises, venous return increases, and the heart briefly slows via the SA node and vagal tone. ^[3,18]3183Balban 202318Spiegel/Stanford 2023 This is why extending the exhale - rather than the inhale - is the most likely active component. The proven lever may be the slow overall rate, not the exhale-to-inhale ratio. ^[23,24]232423Birdee 202324Meehan & Shaffer 2024 Stress favors inhale; relief physiologically favors exhale.

Layer three: the 200-neuron circuit. Two populations of roughly 200 neurons each, located in your brainstem, run every sigh you've ever taken. Neurons in the RTN/pFRG region release neuropeptides called neuromedin B (NMB) and gastrin-releasing peptide (GRP). These signal to the preBötzinger complex - the respiratory rhythm generator - which converts a normal inspiratory burst into a double-inhale burst. Block one peptide and sighing falls by half; block both and sighing stops entirely. Normal breathing continues. ^[7]77Li 2016 The sigh is a separate motor program, not just a bigger breath.

The preBötzinger complex, incidentally, was named after a wine bottle. Jack Feldman's team discovered it at a winery in Bötzingen, Germany, in 1978, and named the region accordingly. The 1991 Science paper that formally identified it is one of the most cited papers in respiratory neuroscience. ^[6]66Smith 1991

There's also a built-in cooldown: after each sigh, the brainstem enforces a refractory period before it will generate another. ^[20]2020Cui 2025 This isn't arbitrary - it's what prevents sighs from cascading into hyperventilation. The pacing instruction ("go slow, don't rush") isn't just safety advice. It's the body's own rule.

And here's a piece of philosophy hidden in the mechanism: you can't voluntarily stop a sigh any more than you can stop a sneeze. The reflex runs on pathways separate from volitional breathing control. Patients who are neurologically locked-in and cannot breathe on command still laugh spontaneously - emotional breathing runs on a different track. ^[19]1919Feldman/Huberman Lab When you do cyclic sighing, you're not imposing a new pattern. You're borrowing the body's own hardware.

The five-minute dose, step by step

Two tools; one mechanism.

Tool one: the daily five-minute practice (RCT-backed). This is the tested dose. Repeat the double inhale / long exhale cycle continuously for five minutes. Effects were measurable after a single session and grew with daily practice over 28 days. The fourth week produced larger benefits than the first. ^[3]33Balban 2023

1. First inhale - Breathe in slowly through your nose. Fill to about 70% of capacity - roughly 3 - 4 seconds. This is important: don't fill all the way yet. Leave room.
2. Second inhale - Without exhaling, take a small additional breath through your nose - a gentle "top-off," not a forceful sniff - roughly 1 - 2 seconds.
3. Long exhale - Breathe out slowly through your mouth until your lungs are fully empty. Six to eight seconds minimum. The slow exhale is what settles the body - don't rush it.
4. Repeat - One complete cycle takes roughly 15 - 30 seconds. Continue for five minutes. Eyes open or closed; seated or lying down.

Tool two: one to three acute sighs (mechanism-plausible, not RCT-tested). In a tense meeting, before a difficult conversation, at the moment of an unexpected setback - one to three slow cyclic sighs. The mechanism supports acute autonomic benefit. This hasn't been tested in a controlled trial at this dose, so the claim is "plausible and practitioner-supported," not "proven." Treat it as the logical smaller dose of something whose larger dose has evidence.

Make it a morning habit, not a crisis tool

The research used one dose: five minutes daily for four weeks. Effects accumulated. Here's what that means in practice.

Start the morning, not the crisis. The RCT ran as a daily morning practice, not an emergency intervention. A practical anchor: before you open your phone, before coffee - five minutes of nothing but the double inhale and the slow exhale. By the time stress arrives, you're not building the habit; you're drawing on it. The fourth week of practice delivered more than the first, which is exactly what physiological adaptation looks like, not just a mood bump.

Showing up is what tracked with the gains. Within the trial, the people who logged more sessions tended to come away with bigger benefits - a pattern, not a proven law of dosing. Still, it points the right way: a short session beats a skipped one. If five minutes feels like too much on a hard morning, two cycles is better than none - the mechanism doesn't need the full dose to run.

One to three sighs in a pinch. A few slow sighs before a difficult conversation, when you catch yourself holding your breath, or the moment a meeting tips sideways borrow directly from the daily practice - the brainstem already knows the pattern. The honest bound, as in the protocol above: this acute dose hasn't been tested on its own, so it rides on the daily practice's evidence rather than its own.

The 28-day window: The only tested protocol runs four weeks. Whether the benefits last beyond that, whether they decay, or whether maintenance doses are needed - all unknown. Treat four weeks as the evidence-backed investment window; what happens after is genuinely uncharted.

Myths and facts

FAQ

Will cyclic sighing help during a real panic attack - or could it make things worse?

If you don't have panic disorder: it may help, with the right form. One to three slow, gentle cycles - exhale-led, not rushed - can shift autonomic state. If it's not working after thirty seconds, stop; don't force it. If you do have panic disorder: the physiology cuts the other way. Panic disorder involves chronically low CO₂ and slow recovery after each sigh; adding deliberate sighs can worsen that imbalance. This isn't a general caution - it's a specific mismatch for a specific physiology. For panic disorder, see breathwork for panic attacks for better-matched options. ^[12]1212Abelson 2001

Am I doing it right? What's the most common mistake?

The most common error is filling your lungs completely on the first inhale, leaving no room for the second breath. When there's no room, the second inhale becomes a forceful sniff - which can cause dizziness or light-headedness if repeated. The fix: treat the first inhale as filling to 70%, not full. The second breath is gentle, a "top-off" - not a gasp. The exhale should be slow and complete; rushing it defeats the mechanism. If you feel dizzy: stop, breathe normally, and try again with a softer first inhale. ^[3]33Balban 2023

How do I build this into my life, and when will I feel results?

The tested dose is five minutes, once daily, for four weeks. Effects were measurable after the first session - small but real. They built over the month; week four showed larger benefits than week one. A morning anchor works well: before checking your phone, before coffee. Acute use (one to three sighs when needed) is mechanism-plausible but not separately RCT-tested at that dose. For habit-building, five minutes is the threshold that has evidence. ^[3]33Balban 2023

Is this better than box breathing, 4-7-8, or mindfulness?

Against mindfulness: cyclic sighing wins on mood, ties on anxiety. Against box breathing and 4-7-8: the RCT included box breathing but was underpowered to prove between-arm differences statistically. Cyclic sighing had the best numerical profile across mood, negative affect, and respiratory rate - but "best profile" isn't the same as statistically proven superiority over each competitor. Use cyclic sighing if you want the technique with the clearest modern evidence base and no breath holds. Use box breathing if you want structured counting. Use 4-7-8 if you want a longer exhale-hold for sleep onset. ^[3,15]3153Balban 202315Fincham 2023 meta

Is the Huberman hype real, or overclaimed?

Partially. The Stanford RCT is real and well-designed - one of the strongest single-technique breathwork studies in the literature. Huberman co-authored it and has publicized it accurately on some metrics. Where the hype outruns the evidence: the "fastest way to calm down" claim (never head-to-head tested at matched latency), the sleep and heart rate improvements (null in the RCT), and the "beats mindfulness on anxiety" framing (mindfulness was marginally better). The technique survives honest scrutiny better than most of what circulates in breathwork spaces. That's the honest position. ^[3,21]3213Balban 202321Huberman newsletter

For professionals

If you want to look behind the curtain, this is the mechanism in full - a denser, more technical register.

The neural circuit (Li et al., Nature 2016). Sighing is generated by a dedicated peptidergic microcircuit in the murine brainstem, separable from the eupneic rhythm. Two populations of ~200 neurons each: RTN/pFRG neurons express neuromedin B (NMB) and gastrin-releasing peptide (GRP); preBötC neurons express the corresponding receptors (NMBR, GRPR) in overlapping subsets. Pharmacological inhibition of either receptor alone reduces sighing by ~50%; blocking both abolishes it. Receptor-expressing neuron ablation eliminates basal and hypoxia-induced sighing while leaving eupneic breathing intact initially, though lung compliance then deteriorates over days. ^[7]77Li 2016 The 2025 Cui et al. follow-on (eLife, PMC12187131) identifies preBötC SST neurons as the downstream common effector; NMB/GRP signaling is sufficient but not obligatory - increased excitability suffices when receptors are blocked. ^[20]2020Cui 2025

The Smith et al. 1991 Science paper formally established the preBötzinger complex as the mammalian respiratory rhythm generator. ^[6]66Smith 1991

The bombesin dose experiment. In preclinical work discussed in the Huberman Lab episode with Feldman, injecting bombesin (the structural family that includes GRP) into the preBötC drove sigh rate from ~25/hour to approximately 500/hour. Ablating the ~50 core SST receptor-expressing neurons eliminated all spontaneous sighing; the animals' lungs began to deteriorate and they required euthanasia within days. The alveolar maintenance hypothesis is not speculative. ^[19]1919Feldman/Huberman Lab

The RSA pathway (Spiegel). Spiegel (Stanford, RCT co-lead) attributes the autonomic calming effect primarily to RSA rather than CO₂ offload: the extended exhale elevates intrathoracic pressure, increases venous return, and activates vagal brake via the SA node. This is the more defensible mechanistic claim - and it directly addresses the Buteyko critique that deliberate sighing depletes CO₂. The calming signal is in the exhale's cardiovascular effect, not in the ventilatory change per se. ^[18,3]18318Spiegel/Stanford 20233Balban 2023

The dual-pathway model (Feldman). Volitional breathing - counting breaths in meditation, performing a yoga pranayama - runs on a corticospinal tract overlying the autonomic breathing circuit. Emotional breathing (laughing, crying, sighing involuntarily) runs on a separate subcortical track. Locked-in patients who have lost volitional breath control still laugh and cry. When you deliberately perform the cyclic sigh, you are engaging the volitional pathway to trigger the subcortical one - borrowing the body's own effector. ^[19]1919Feldman/Huberman Lab

The psychophysiology reset model (Vlemincx 2009). Spontaneous sighs occur approximately 12 times per hour at baseline, clustered preferentially at relief - the moment of safety-signal, not peak stress. ^[10]1010Vlemincx 2009 In rats, a ~20-fold rise in sigh rate at relief (vs quiet rest between trials) has been recorded - rat data, not a human figure. ^[11]1111Soltysik & Jelen 2005 (rats) The voluntary technique may work partly by mimicking this relief-signature to the brainstem - signaling "threat resolved" before the cognitive context changes. This remains mechanistically plausible, not RCT-confirmed at the level of the signaling chain.

The Abelson 2001 panic-physiology data. Sixteen panic disorder patients showed: elevated spontaneous sigh frequency vs controls; lower end-tidal pCO₂ (chronic hypocapnia); and slower pCO₂ recovery after sighs compared to GAD patients and controls. Sighing in PD patients is not compensatory (it is not triggered by low tidal volume or CO₂ elevation) - it appears to be a non-homeostatic amplification of the reflex. Deliberate cyclic sighing sessions have not been tested in this population; the theoretical concern is additive hypocapnia in an already-hypocapnic system. The recommendation against cyclic sighing in panic disorder is a clinical inference from physiology, not a direct contraindication from a trial. ^[12]1212Abelson 2001

Waterloo / Boucher 2024 poster. A non-peer-reviewed conference poster (Psychonomic Society, 2024) found HRV increases with cyclic sighing but null mood effects. This is cite-level corroboration for the body-calm finding, consistent with the body/mood split. Do not treat it as independent replication: non-peer-reviewed, not published. ^[22]2222Boucher 2024 poster

Evidence summary for clinical context. The Balban RCT (N=108, retrospective registration NCT05304000) is the primary human evidence. Its limitations: healthy adults only; moderate-to-severe psychiatric conditions excluded; remote self-report adherence; between-arm comparisons explicitly underpowered; four-week follow-up only. The Hanley 2025 pain pilot (J Behav Med, 10.1007/s10865-024-00548-5; N=81, randomized 1:1, orthopedic waiting room) provides independent evidence for the pain application, with anxiety/depression null - consistent with the body-calm > mood-lift hierarchy. NCT07379606 is the replication trial (cognition outcomes, exercise comparator; not yet published). Grade: moderate. Pre-registered primary hypothesis separating body-calm from mood, a sham control arm, and a clinical-population cohort are the three design gaps the next trial needs to close. ^[3,13]3133Balban 202313Hanley 2025