Decreased CO2 saturation during circular breathwork supports emergence of altered states of consciousness

Abstract

Altered states of consciousness (ASCs), induced e.g. during psychedelic-assisted therapy, show potential to treat prevalent mental health disorders like depression and posttraumatic stress disorder. However, access to such treatments is restricted by legal, medical, and financial barriers. Circular breathwork may present a non-pharmacological and hence more accessible alternative to engage similar therapeutic processes. Scientific studies of breathwork are only just emerging and its physiological and psychological mechanisms are largely unknown. Here, we track physiological and experiential dynamics throughout a breathwork session, comparing two forms of breathwork: Holotropic and Conscious-Connected breathwork. We show that a reduction in end-tidal CO₂ pressure due to deliberate hyperventilation is significantly correlated to ASC onset (r = -0.46; p < 0.001). Based on standard questionnaires (MEQ-30 and 11-DASC), the ASCs evoked by breathwork resembled those produced by psychedelics across several experiential domains such as ego dissolution, and their depth predicted psychological and physiological follow-on effects, including improved well-being and reduced depressive symptoms. Further analysis showed that different breathwork approaches produced highly similar outcomes. Our findings identify physiological boundary conditions for ASCs to arise in a non-pharmacological context, shedding light on the functional mechanisms of breathwork as well as its potential as a psychotherapeutic tool.

Fig. 1. End-tidal CO₂ saturation during breathwork.

A Boxplot of mean partial end-tidal CO₂ pressure (etCO₂) per participant, pooled across the active portion of the breathwork session (time points 2–4 out of 6). Central line: Median. Box outline: 25th and 75th percentile. Whiskers: 10th and 90th percentile. Blue: Holotropic (n = 20 participants for active-breath group, n = 10 for passive-breath group). Red: Conscious-Connected (n = 23 participants for active-breath group, n = 8 for passive-breath group). Desaturated colours (on the left): Passive-breath control groups. Saturated colours (on the right): Active-breath groups. Both active-breath groups showed significantly lower etCO₂ than their passive-breath counterparts. B Same as A for minimum etCO₂ per participant across measurement time points 2–4. Differences between active and passive breath groups are even more pronounced, with active-breath participants reaching etCO₂ as low as 10–20 mmHg throughout the session. C Time course of etCO₂ across all six measurement time points. Error bars: SEM. Blue: Holotropic (n = 20 participants for active-breath group, n = 10 for passive-breath group). Red: Conscious-Connected (n = 23 participants for active-breath group, n = 8 for passive-breath group). etCO₂ decreased rapidly at the start of the session, reaching their minimum around measurement time points 2–3, and then increasing gradually towards the end of the session. Note that for Conscious-Connected breathwork sessions, measurements were taken approx. every 15 min, while for Holotropic breathwork sessions, measurements were taken approx. every 30 min (see “Methods” and Table 1).

Fig. 2. Experience depth during breathwork.

A Subjective experience triggered by breathwork, as assessed by the 11-Dimensional Altered States of Consciousness Scale (11D-ASC). Dark purple: Active-breath condition (n = 43 participants). Light purple: Passive-breath condition (n = 18 participants). Dashed lines: Typical 11D-ASC scores for three typical psychedelic treatments (see in-figure legend), with data extracted from the Altered States of Consciousness Database (see “Methods” and Supplementary Table S4): 20–25 mg Psilocybin (oral; scores pooled across five relevant clinical studies; see Supplementary Table S4), 0.1 mg LSD oral/0.075 mg intravenous (pooled across six studies), and 0.125 mg MDMA (oral; pooled across five studies) .Solid grey line: Typical 11D-ASC scores for placebo treatments (three clinical studies, see Supplementary Table S4). B Same as (A) for the four subscales of the Mystical Experiences Questionnaire 30 (MEQ30). Reference studies for psychedelic and placebo treatments are listed in Supplementary Table S5. C Boxplot of mean experience depth per participant, determined by hand signs given on a scale of 1–5 (1= every-day consciousness, 5 = deeply altered consciousness; see “Methods”), which were then pooled across measurement time points 2–4. Central line: Median. Box outline: 25th and 75th percentile. Whiskers: 10th and 90th percentile. Colour scheme as in (A, B). Holotropic: n = 20 participants for active-breath, n = 10 for passive-breath. Conscious-Connected: n = 23 participants for active-breath, n = 8 for passive-breath. Active-breath groups on average indicated deeper experiences than their passive-breath counter parts. D Same as (C) for the maximum experience depth. E Same time course as shown in Fig. 1C, but for experience depth. Holotropic: n = 20 participants for active-breath, n = 10 for passive-breath. Conscious-Connected: n = 23 participants for active-breath, n = 8 for passive-breath. In both active-breath groups, consciousness is increasingly altered at the start of the session, reaching its maximum around measurement time points 3–4, and then gradually reverts to baseline. A qualitatively similar but quantitatively shallower dynamic can be seen in both passive-breath groups. Note that for Conscious-Connected breathwork sessions measurements were taken approx. every 15 min, while for Holotropic breathwork sessions, measurements were taken approx. every 30 min (see “Methods” and Table 1).

Fig. 3. Relationship between end-tidal CO₂ saturation and experience depth.

A Regression between etCO₂ and experience depth, separated into six measurement time points. Dark purple: Active-breath condition (n = 43 participants). Light purple: Passive-breath condition (n = 18 participants). B Scatter plot showing mean experience depth per participant, averaged across measurement time points 2–4, as a function of mean end-tidal CO₂ pressure (etCO₂) averaged across the same interval (Holotropic: n = 30 participants; Conscious-Connected: n = 31 participants). C Average trajectory of etCO₂ and experience depth throughout a session of Holotropic breathwork. Filled circles: Average etCO₂ and experience depth at each time point, averaged across participants of the active-breath group. Thin horizontal and vertical lines: SEM for etCO₂ and experience depth, respectively, for each time point (n = 30 participants). Darker colours signify measurement time points at the beginning of the session, light colours denote time points towards the end of the session. Open circles: Same for the passive-breath control group. D Same as (B) for Conscious-Connected breathwork (n = 31 participants).

Fig. 4. Psychological effects of breathwork.

A Scores of the Self-Report Quick Inventory of Depressive Symptomatology (QIDS-SR16) one week before and after the breathwork session, completed by a total of 20 participants in the active-breath group. Red lines: Participants in Conscious-Connected breathwork (n = 12). Blue lines: Participants in Holotropic breathwork (n = 8). Black line and error bars: Mean ± SEM across participants. Star indicates statistically significant decrease in QIDS-SR16 scores at an Alpha level of 0.05. B Same as (A) for the passive-breath group that completed follow-on questionnaires. Light red: Conscious-Connected (n = 3 participants). Light blue: Holotropic (n = 2 participants). Grey line and error bars: Mean ± SEM. C Same as (A) for scores in the Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS). Participant numbers as in (A). Star indicates statistically significant increase in WEMWBS scores at an Alpha level of 0.05. D Same as (B) for WEMWBS scores, including the same participant numbers. E Correlations between post-breathwork changes in QIDS-SR16 and WEMWBS scores on the one hand, and acute aspects of the breathwork session on the other hand, specifically average etCO₂ during the active part of the session, as well as scores in the Mystical Experiences Questionnaire 30 (MEQ30) and the three subscales of the 11-Dimensional Altered States of Consciousness scale (11-DASC), which are termed ‘Oceanic boundlessness’, ‘Ego dissolution’ and ‘Visual reconstruction’. Correlation strength is colour-coded (see in-figure legend). Stars indicate statistically significant correlations. Correlations were computed based on all 25 participants who completed questionnaires one week before and after the breathwork session. Post-breathwork changes in QIDS-SR16 scores were not significantly predicted by any of the acute parameters of the breathwork session. In contrast, improvements in WEMWBS scores were predicted by lower etCO₂ and by deeper ASCs during the session, indicated by both MEQ30 and 11-DASC scores.

Fig. 5. Subacute physiological effects of breathwork.

A Concentration of the autonomic nervous system activity marker alpha-amylase (α-amylase) directly before and after the breathwork session, completed by a total of 40 participants in the active-breath group. Red lines: Participants in Conscious-Connected breathwork. Blue lines: Participants in Holotropic breathwork. Black line and error bars: Mean ± SEM across participants. Star indicates statistically significant decrease in α-amylase at an Alpha level of 0.05. B Same as (A) for passive-breath participants (n = 15). Light red: Conscious-Connected. Light blue: Holotropic. Grey line and error bars: Mean ± SEM. C Same as A for concentrations of the inflammatory marker interleukin-1 beta (IL-1β). Star indicates statistically significant increase in IL-1β at an Alpha level of 0.05. D Same as B for concentrations of IL-1β. Star indicates statistically significant increase in IL-1β at an Alpha level of 0.05. E Same as Fig. 4E for correlations between the acute aspects of the breathwork session and the post-breathwork changes in α-amylase and IL-1β levels. IL-1β increases were linked to subjective experiences, but not etCO₂, during breathwork, such that deeper ASCs predicted smaller increases in IL-1β post-session. Change in α-amylase was not significantly predicted by any of the acute parameters of the breathwork session.