The research, and its limits

Human Again draws on research into stress physiology, recovery and respiratory function. That research informs the approach. It does not establish the effectiveness of the complete program, and this page says which is which.

The page is ordered from the better-evidenced to the less. The autonomic effects of slow breathing rest on meta-analyzed evidence. The claim that training tolerance to carbon dioxide produces benefits in healthy working adults does not, and the section on the evidence says so in plain terms. Each section closes with a paragraph marked Where this stops, which states what its sources do not establish.

Our biology and our conditions are out of step

Biological evolution moves far more slowly than sociocultural change (Comer & Powers, 2024). When a trait that served an ancestral environment becomes less suited to a rapidly changed one, researchers call it evolutionary mismatch (Li et al., 2018).

The stress response is the example that matters here. Rapid mobilization once improved our chances of surviving a threat: heart rate may rise, breathing may become faster and shallower, pupils may widen and you become more alert, while digestion becomes less of a priority. The same biology now responds to demands that may be social, cognitive, digital or almost continuous — deadlines, traffic, information overload, availability, persistent performance pressure.

What that design assumes is a challenge followed by a recovery period. Continuous demand leaves too little room for the second half.

Psychological stress involves emotional and physical responses to demands we experience as threatening or overwhelming (American Psychological Association, 2018). How we interpret those demands matters too: does this feel manageable, or does it seem to ask more of me than I currently have to give? The transactional model of stress and coping describes this interaction between the person and their environment (Lazarus & Folkman, 1984; Obbarius et al., 2021). It helps explain why two people can respond differently to the same event, and why your own response can change over time.

The familiar split between good and bad stress is less settled than it sounds. You may encounter eustress, for demands that support growth or performance, and distress, for demands experienced as harmful or unmanageable. Bienertova-Vasku and colleagues (2020) argue the two are not biologically distinct, and that the difference lies in how a demand is appraised and how it turns out rather than in the stress response itself.

Duration matters as well. Acute stress is brief and linked to an immediate demand, and our biology can often manage that mobilization well, especially when recovery follows. Tolerable stress may involve difficult work or performance demands that stay manageable because you have enough resources, support and ways to cope. But manageable does not mean cost-free indefinitely: even tolerable pressure can become costly when recovery keeps falling short.

Where this stops. Evolutionary mismatch is a hypothesis, and that is how the cited work frames it (Li et al., 2018). It organizes these observations coherently. It does not predict what any individual will experience.

The cost accumulates where it is hardest to see

Your body works continually to keep conditions such as temperature, blood pH and hormonal activity within a workable range. That is homeostasis. Maintaining it takes constant adjustment: your brain and body anticipate needs and respond to physical, psychological and environmental demands, with the autonomic nervous system, the HPA axis, the immune system and other regulatory networks all taking part. This is allostasis — achieving stability through change.

When those systems are activated repeatedly, remain active too long, or regulate inefficiently, a cumulative burden can develop. This is allostatic load. The processes that help protect stability can themselves become a source of strain (McEwen, 2017).

Workload and sustained pressure contribute. So do constant availability, noise and light exposure, air pollution and sheer volume of information, alongside limited movement and limited recovery.

Familiarity is what makes this hard to notice. Think of a washing machine running while you work from home: after a while you stop registering it, and only when it stops do you hear how quiet the room is. Getting used to a demand does not mean the body has stopped responding to it.

It is also why time away from work does not reliably restore. Being inactive and recovering well are different things — you can stop working and still be taking in a great deal.

Recovery research treats this as a subject of its own (Sonnentag et al., 2022). It separates experiences such as mentally disengaging from work, relaxation, positively challenging activities and choice over free time; their relationships with well-being and work outcomes differ (Sonnentag & Fritz, 2007; Headrick et al., 2023). Studies also suggest that psychological detachment can improve through structured interventions (Karabinski et al., 2021).

Where this stops. Allostasis and allostatic load describe how adaptation works and how it can become costly. They are models, not a measurement I can give you about yourself. And the detachment research tested interventions of its own, not this one: it supports the claim that recovery is trainable, not the claim that this program trains it.

Where that cost shows up

I work from four connected dimensions. They are my own synthesis of the biopsychosocial perspective and of well-being research — a working frame, not a validated instrument.

The four influence one another continually, and what you need within each changes with your circumstances. Context matters too: you may feel highly capable at work while having much less control elsewhere, or the reverse.

Sustained stress may contribute to irritability, anxiety, low mood, repetitive thinking or rumination, disturbed sleep and reduced concentration. The table below sets out the associations that carry sources, and names the kind of stress each body of work examined — they are not all the same.

Responding well takes three things: recognizing what you need, having enough capacity to adapt, and having practical ways to adjust your thinking and behavior. Noticing a need and being able to act on it are separate abilities. The program returns to the first of them as interoceptive awareness — the capacity to notice and interpret signals from inside your own body.

Where this stops. Every entry in that table is an association rather than a mechanism demonstrated in an individual. These associations depend on the intensity and duration of stress and on the person’s wider health context, and how sustained stress shows up can differ across people, situations and areas of life. High perceived stress has been associated with a moderately higher incidence of coronary heart disease (Richardson et al., 2012) — an association across populations, not a prediction about you, and not evidence that this program prevents cardiovascular disease.

Why the work starts with breathing

Many approaches begin by changing how you think or what you do, and they can be highly effective. They also become harder to act on the longer you have been under pressure. A useful idea still takes capacity to use, which is why this program starts with the capacity rather than the idea.

The target is autonomic flexibility: the ability to move between sympathetic mobilization and parasympathetic recovery as the situation requires. It supports internal balance and is one component of resilience.

Breathing is where that system is reachable. Under pressure, even mild sympathetic activation may make breathing faster, shallower, harder, or larger in volume; more movement may occur in the upper chest, and more air may pass through the mouth. When manageable pressure persists, those changes can be repeated often enough to become habitual.

That pattern can close a loop. When breathing volume exceeds current metabolic needs, it may affect carbon dioxide balance: excess ventilation lowers CO₂, which can reduce blood flow to the brain and influence neural activity. Concentration, memory, clarity and emotional regulation may become more difficult, and those changes can then affect thoughts, behavior and sensations in ways that reinforce the experience of stress.

Breathing is the one part of this system that runs automatically and stays open to deliberate influence. That is the argument for beginning there: an argument about access, not about breathing mattering more than anything else.

Functional breathing is assessed across three interacting dimensions — biochemical, biomechanical and psychophysiological (Courtney, 2017). The program works through them in that order, with four principles:

• Light — volume matched to current metabolic need
• Slow — a rate that is calm and appropriate to the situation
• Deep — efficient movement of the diaphragm and lower ribs, not a larger breath
• Nasal — the nose as the usual route whenever circumstances allow

Two of those principles can pull against each other, which is why the order matters. The total air moved each minute is breathing rate multiplied by volume per breath, so these two patterns move the same amount:

• 12 breaths per minute × 500 milliliters per breath = 6 liters per minute
• 6 breaths per minute × 1,000 milliliters per breath = 6 liters per minute

The second is slower, and each breath is twice as large. If a longer exhalation creates air hunger and you compensate with a fuller inhalation, slowing down can leave total ventilation unchanged or higher. This is why the program addresses breathing chemistry before it places greater emphasis on slowing, and why familiar approaches such as 4-7-8 breathing, a 4-6 rhythm, or breathing near 5.5 breaths per minute can produce different results depending on breath size and your current state.

Where challenge is introduced, it is dosed. A small, controlled dose of a stressor can stimulate useful adaptation while a larger dose of the same stressor may be harmful — the principle of hormesis (Wan et al., 2024). Reduced breathing volume and short breath holds are used in exactly that way, and the dose has to fit your present capacity.

This is also why I am cautious with intensive practices built on deliberate hyperventilation. They can produce sensations close to those of panic. Someone already anxious may read those sensations as danger, and someone who finds internal signals hard to interpret may not notice when a practice has become too intense.

One distinction defines the work. A breathing exercise is time-limited and may change how you feel now. Respiratory retraining aims at the pattern that continues outside practice — route, volume, rate and mechanics — so that a more functional pattern gradually becomes automatic. Breathing is partly learned behavior, which is what makes retraining possible at all.

What the evidence supports, and how strongly

The four principles are not equally well evidenced, and it would be easy to write this page as though they were. They are set out below strongest first.

Slow breathing: meta-analyzed

Voluntary slow breathing increases vagally mediated heart-rate variability — during the practice, immediately afterward, and after repeated practice. That comes from a systematic review and meta-analysis (Laborde et al., 2022). An earlier systematic review linked slow breathing with heart-rate variability, respiratory sinus arrhythmia and measures of brain activity, and reported associations with reductions in symptoms of anxiety and low mood, as well as arousal and anger (Zaccaro et al., 2018).

A 2025 narrative review of thirty studies on slow, nasal and diaphragmatic breathing reports a similar pattern: improvement in at least one HRV parameter in twenty-three of them, reduced anxiety in ten, and reduced perceived stress in nine (Little, 2025). Two things about that source belong here rather than in a footnote. It is a narrative review, which carries less weight than a meta-analysis. And its author developed the breathing protocol the paper argues for — a protocol the paper itself calls unvalidated, with no study yet testing it.

The mechanism is well described. As you inhale, the vagus nerve’s influence on the heart usually decreases slightly and heart rate rises; as you exhale, that influence generally becomes stronger and heart rate slows, which helps explain why a calm, unforced exhalation can feel settling. At the heart, vagal nerve endings release acetylcholine, a chemical messenger that slows the heart’s natural pacemaker (Gordan et al., 2015). Breathing also has a mechanical influence: movement of the diaphragm changes pressure inside the chest and abdomen, as well as the sensory information returning to the brain (Russo et al., 2017). At certain rates, which vary between individuals, heart-rate and blood-pressure rhythms move closer to the cardiovascular system’s resonance frequency — the rate at which they reinforce one another most strongly (Lehrer & Gevirtz, 2014).

Where this stops. That same review reports the trials that found nothing. In a randomized controlled trial of seventy-five healthy adults over ten weeks, neither slow breathing nor biofeedback improved HRV or stress scores against a control group. In another, anxiety fell in the breathing group and the comparison group alike, with no difference between them. In a third, participants using a breathing app reported less subjective improvement than those breathing naturally (Little, 2025). The review attributes much of this spread to how often people practiced and how well they were taught.

The outcome measure needs care too. HRV is influenced by age, genetics, hormones and medication, and by sleep, stress, exercise and recovery, breathing, caffeine, meals, posture and the conditions in which it is measured (Uusitalo et al., 2007). It is most useful when measurements are consistent and interpreted in the context of the person and the situation.

A reliable change in an autonomic measure is also not, by itself, a reliable change in how a demanding week feels.

Light breathing and CO₂ tolerance: mechanism established, outcomes not

The physiology of carbon dioxide is not in question. Arterial blood pH normally stays within a narrow range of approximately 7.35–⁠7.45 (Brinkman & Sharma, 2023), and CO₂ is part of the bicarbonate buffering system that helps maintain it. As CO₂ rises locally, pH falls slightly and hemoglobin holds oxygen less tightly, allowing it to be released more readily into active tissue (Patel et al., 2023); this relationship is the Bohr effect, described by Christian Bohr in 1904 (Jensen, 2004). Chemoreceptors detect changes involving CO₂, oxygen and pH, and help determine when the urge to breathe becomes noticeable (Patel et al., 2022).

Two patterns are worth separating. Acute hyperventilation is a rapid increase in ventilation that lowers CO₂. Chronic overbreathing is much less noticeable: habitually moving more air than current energy use requires, through breaths that may be too frequent, too large, or both. Habitual overbreathing may make the chemoreceptor system more sensitive, so that a smaller buildup of CO₂ begins to feel uncomfortable. Raising tolerance for ordinary CO₂ changes is what this part of the program works on.

Where this stops. Konstantin Buteyko developed his method during the 1950s after observing that people who were more unwell often breathed faster and more heavily, drawing on clinical observations and work with a large number of patients. One component of it is addressed in the review above: a brief post-exhalation hold raises end-tidal CO₂ and activates central chemoreceptors, which the review describes as producing a parasympathetic response and a resetting of the chemoreflex toward slower breathing (Little, 2025). The principle of light breathing — habitually reducing breathing volume — has no such support, in that review or in anything else cited here. The step from established CO₂ physiology to a program built partly on reducing volume is a clinical inference I am making. It is not one the literature cited here has made for me.

Nasal breathing: part settled, part developing

Some of it is straightforward. The nose filters, warms and humidifies air before it reaches the lungs, and its natural resistance can help slow airflow and support lighter breathing and more effective diaphragm movement. Nitric oxide is produced in the paranasal sinuses and contributes to widening blood vessels, immune defense and local antimicrobial activity; nasal breathing carries it into the respiratory tract, where it may support pulmonary circulation and oxygen uptake (Lundberg, 2008). In healthy adults, breathing out through the mouth produced about 42 percent more net water loss than breathing out through the nose (Svensson et al., 2006).

The neuroscience is newer. Natural nasal breathing has been found to synchronize electrical activity in the piriform cortex, amygdala and hippocampus, with stronger effects during inhalation and weaker ones when breathing was redirected through the mouth (Zelano et al., 2016). When air bypasses the nose that sensory input is absent, and research suggests this can change breathing-related rhythmic activity and coordination between brain regions (Lee et al., 2020).

Where this stops. Read those studies in their settings. Kuroishi et al. (2015) studied children with mouth-breathing syndrome. Salimi et al. (2022) delivered nasal air puffs to comatose, mechanically ventilated patients as a form of noninvasive brain stimulation. Neither describes an adult at a desk choosing to breathe through the nose. The field is promising and unfinished.

One question of my own, marked as such

Breathing interacts with the baroreflex — the rapid feedback system that regulates blood pressure — by acting as a rhythmic stimulus to the cardiovascular system, and the timing between breathing, heart rate and blood pressure shifts with breathing rate (Lehrer & Gevirtz, 2014). The role of breathing volume in that interaction was not researched. Whether breathing that lowers CO₂ chemosensitivity is associated with higher baroreflex sensitivity, and so contributes to homeostatic blood-pressure levels, is a question I find plausible and cannot cite. I include it because it sits underneath the approach, not because it is established.

What this doesn’t establish yet

No study tests this program. Everything above supports its components — not the nine-week sequence they are used in, and not the outcomes I describe for working professionals.

The timeframes I give you come from practice rather than research. Breathing educators sometimes use approximately six weeks as the point at which nasal breathing can begin to feel more automatic. Within breathing education, around twelve weeks is sometimes suggested as a timeframe for substantial adaptation; that estimate comes from practical experience, and research has not clearly established it as a reliable timeline. Individual progress varies.

Two claims I make elsewhere on this site I describe as clinical observation rather than cite, because that is what they are: that prolonged stress can change how a person breathes by default, and that cognitive and behavioral change costs more and holds less well under continuous demand.

I expect some of this to change. Testing the program properly is work I intend to do with university partners.

Scope. This is psychological consulting and psychoeducation. It addresses stress regulation, resilience, well-being and sustainable performance. It is not psychotherapy, and it does not replace medical or mental health treatment.