Introduction

Interoception encompasses the complex neurocognitive processes of perceiving, integrating, interpreting, and regulating internal bodily signals, serving as a critical substrate for physiological homeostasis1. This system operates through bidirectional neural pathways: Ascending (afferent) pathways transmit visceral signals from peripheral organs (e.g., cardiac, gastric, and vesical activity) to central processing regions, notably the insular cortex; descending (efferent) pathways mediate top-down modulation of autonomic and physiological responses. Grounded in this framework, Garfinkel et al.2 established a tripartite model comprising three interlinked yet functionally distinct components: interoceptive accuracy (IAc)—the objective ability to detect physiological fluctuations, typically quantified through heartbeat perception tasks (HPTs) such as the heartbeat tracking method (counting perceived heartbeats without external feedback)3 or the heartbeat discrimination paradigm (detecting temporal congruency between cardiac cycles and external stimuli)4; interoceptive sensitivity (IS)—subjective appraisal of bodily awareness measured via standardized self-report instruments such as the Multidimensional Assessment of Interoceptive Awareness (MAIA), a multidimensional questionnaire that captures different facets of body awareness5; and interoceptive awareness (IAw)—the metacognitive capacity to evaluate interoceptive performance, operationalized as the discrepancy between IAc and IS6.

Accumulating evidence delineates interoception as a pivotal regulator across multiple psychological domains, including emotional processing7, decision-making8, behavioral prediction9, and the regulation of physical depletion10. Modern models of sports regulation indicate that interoception contributes to the perception of bodily regulation11. Interoception also monitors the body’s physiological arousal levels through both top-down and bottom-up pathways, ensuring homeostasis and facilitating the integration of early sensory stimuli with subsequent emotional responses12. Clinical investigations further reveal that interoceptive dysfunction constitutes a transdiagnostic mechanism underlying diverse neurological and psychiatric conditions. Structural and functional perturbations within interoceptive neural networks—including altered insular connectivity, aberrant thalamocortical signaling, and vagal afferent dysregulation—have been implicated in conditions spanning chronic pain syndromes13, substance use disorders14, affective disorders15, autism spectrum disorder16, and neurodegenerative diseases17. These neurobiological alterations may drive the high comorbidity observed between mental health disorders and somatic pathologies18.

The critical role of interoception in health maintenance has spurred concerted efforts to develop effective interventions, with current strategies broadly classified into pharmacological, neurostimulation, and behavioral domains. Pharmacological approaches targeting neurotransmitter systems or peripheral interoceptors demonstrate modest efficacy in laboratory settings but face translational challenges due to systemic side effects and poor target specificity19. While invasive neuromodulation techniques like deep brain stimulation show potential for recalibrating central interoceptive networks, their clinical utility remains constrained by anatomical localization uncertainties and procedural risks20. In contrast, behavioral interventions exhibit enhanced safety profiles alongside measurable improvements in IAc, likely mediated through somatosensory attentional mechanisms21. Converging findings from sport science and research on mind-body interventions further indicate that interoceptive processes are malleable in both expert and novice populations. In elite performers, long-term sensorimotor training across a range of sports has been associated with higher levels of interoceptive accuracy, sensitivity, and awareness compared with recreational athletes and non-athletes22. Expert dancers likewise display superior heartbeat perception relative to non-dancers, suggesting that highly specialized movement training can refine IAc over years of practice23. Parallel work on yoga, mindfulness and clinical interventions has begun to quantify interoceptive outcomes more directly. A twelve-week Hatha yoga programme in people with multiple sclerosis increased body awareness as measured by the MAIA, particularly in dimensions related to attentional regulation, emotional awareness and trust in bodily sensations24, whereas a ten-week Hatha yoga course in healthy university students did not alter cardiac or gastric IAc, indicating stronger effects on subjective than on objective indices of interoception25. Short mindfulness-based programmes similarly produce reliable improvements in self-reported IS and reductions in state anxiety without consistently changing performance on heartbeat tracking tasks26. Clinically focused interventions such as the Mindful Awareness in Body-oriented Therapy intervention and the Aligning Dimensions of Interoceptive Experience protocol explicitly train interoceptive skills in women receiving treatment for substance use disorders and in autistic adults, respectively, and have been linked to improvements in interoceptive awareness and reductions in anxiety-related symptoms27,28. Collectively, these lines of evidence indicate that interoception is not a fixed trait but a capacity that can be modified through sustained practice in both high-performance and therapeutic contexts.

Despite this progress, important gaps remain in the interoceptive training literature. Empirical work to date has largely focused on Western sports and dance, or on seated yoga and mindfulness-based courses delivered in clinical or educational settings. By contrast, Chinese traditional mind-body practices such as Tai Chi, Baduanjin and standing meditation forms have received relatively little systematic investigation, even though they explicitly integrate posture, breath regulation and prolonged internally directed attention within a coherent training system. Within the broader landscape of interoceptive training, these practices are therefore likely to constitute powerful yet underexplored approaches to shaping interoception across multiple dimensions. Against this backdrop, the present study focuses on Hunyuan Zhuang, a foundational standing technique in Chinese traditional physical culture. In Hunyuan Zhuang, practitioners maintain a single, expansive posture for prolonged periods while coordinating diaphragmatic breathing and focused mental intention. The practice embodies the philosophical principle that “large movements are inferior to small movements, small movements are inferior to stillness, and the movement within stillness is the source of unceasing vitality”29, emphasizing the cultivation of internal awareness amidst external quiescence. From a contemporary theoretical standpoint, Hunyuan Zhuang can be interpreted within predictive coding accounts of interoception, which hold that the brain continuously generates top-down predictions about bodily states and updates these predictions by minimizing mismatches with ascending sensory signals. By orchestrating posture, breathing and internally directed attention in a sustained and structured manner, Hunyuan Zhuang may alter both the precision of bottom-up interoceptive signals and the form of top-down interoceptive predictions, thereby offering a distinctive, culturally grounded route for interoceptive training.

In line with this account, the interoceptive regulatory mechanisms of Hunyuan Zhuang can be conceptualized through several interacting components. First, its characteristic expansive posture, which maintains a delicate balance between isometric tension and relaxation, is hypothesized to provide a unique somatic signal. This aligns with empirical evidence demonstrating that expansive postural configurations can reliably increase self-reported feelings of power and other positive affective states relative to contractive displays30, likely by optimizing afferent input to the central nervous system. Second, the practice employs volitional, diaphragmatic breathing synchronized with the postural framework. This is of critical importance, as slow, paced respiration has been shown to enhance parasympathetic (vagal) activity and support emotional regulation and psychological well-being, effects that have been hypothesized to arise in part from voluntary regulation of internal bodily states31. Third, the core element of directed mental focus, or intention, guides attention to specific internal bodily loci. This process is analogous to interoceptive attention training in mindfulness, which has been neurophysiologically linked to the refinement of sensory acuity and the enhanced cortical representation of internal bodily states32. Collectively, the simultaneous engagement of these components suggests that Hunyuan Zhuang may facilitate a state of heightened bodily awareness by both amplifying bottom-up interoceptive signals and refining top-down attentional filters, a combined mechanism highlighted in contemporary accounts of body-centred contemplative training33.

Hunyuan Zhuang, characterized as an open-posture standing practice, shares fundamental characteristics with expansive postures established in prior research. Postural configurations are operationally defined as geometric relationships between two or more body components (e.g., upper limbs and trunk), quantitatively determined through joint angles across bodily segments34. Current investigations in embodied cognition have systematically examined how postural dynamics modulate psychological states, with particular emphasis on differential effects between expansive (limb-extension) and contractive (limb-flexion) configurations. Critical methodological insights from recent meta-analytic evidence emphasize the necessity of incorporating neutral posture controls to differentiate valence-specific outcomes - distinguishing whether observed effects originate from beneficial expansive postures, detrimental contractive postures, or their interactive combinations30. Aligned with this experimental paradigm, the present study employs a triadic comparative design contrasting Hunyuan Zhuang’s open-posture characteristics against both contractive and neutral configurations to elucidate its unique psychological effects through systematic comparison.

While extant literature substantiates Hunyuan Zhuang’s potential for interoceptive enhancement through phenomenological reports, rigorous empirical validation remains scarce. This investigation systematically examines its psychophysiological mechanisms through two complementary studies: Study 1 adopts an expert-novice comparative paradigm to assess acute interoceptive advantages relative to contractive/neutral postures; Study 2 implements a randomized controlled trial evaluating longitudinal effects following an 8-week intervention. By elucidating these mechanisms, our research not only advances non-pharmacological interoceptive intervention strategies but also contributes to the scientization of traditional Chinese martial arts. This dual contribution supports the integration of ancient cultivation practices into contemporary neurocognitive frameworks while preserving their cultural essence.

Experiment 1: the effects of acute hunyuan zhuang practice on interoception

This study employs an expert-novice paradigm to address two aims. First, to determine whether long-term practitioners of Hunyuan Zhuang exhibit superior IAc and IS compared to novices. Second, to contrast the acute effects of adopting the Hunyuan Zhuang posture against a contractive and a neutral control posture on the immediate perception of internal physiological signals.

Methods

Participants

The sample size was determined a priori using G*Power 3.1 for a repeated-measures analysis of variance (ANOVA) with a within-between interaction. Based on an anticipated medium effect size (f = 0.25, corresponding to η² ≈ 0.06), an α level of 0.05, and a desired statistical power (1 - β) of 0.80, the analysis indicated a minimum required total sample size of 28 participants. To accommodate potential attrition, we recruited a larger cohort. Participants were recruited via open advertisements from two institutions in Beijing: China University of Geosciences and the Chinese Wushu Academy at Beijing Sport University. All volunteers were screened against a set of common eligibility criteria before group assignment. Common inclusion criteria required participants to be (1) between 18 and 30 years of age, and (2) right-handed. Common exclusion criteria were: (1) any known cardiovascular, respiratory, neurological, or psychiatric disorder; (2) any musculoskeletal condition precluding the maintenance of standing postures for 10 min; or (3) regular use of medication known to affect heart rate or cognitive function (e.g., beta-blockers, stimulants). Eligible participants were then assigned to one of two cohorts based on group-specific criteria: Expert Group – defined by three stringent criteria: (i) a minimum of 6 years of continuous martial arts training, (ii) at least 3 years of specialized training in Hunyuan Zhuang, and (iii) a maintained practice schedule of at least three 60-minute sessions per week over the preceding three years. Novice Group – defined by the absence of any prior systematic training in mind-body practices (e.g., yoga, Tai Chi, Qigong, Zen or seated meditation) or any form of martial arts. To ensure group purity, individuals reporting any prior experience in these practices were excluded from the novice group. Initial recruitment yielded 66 eligible volunteers. Nine participants discontinued due to scheduling conflicts during testing (attrition rate: 13.6%), resulting in 57 complete datasets for analysis (novices: 19 M/11F, n = 30; experts: 16 M/11F, n = 27). Comprehensive demographic characteristics are presented in Table 1.

Table 1 Baseline demographic information of participants (M ± SD).
Full size table

Experimental design

A 2 (Group: Expert, Novice) × 3 (Posture: Hunyuan Zhuang, Contractive Posture, Neutral Posture) mixed-design was employed. The between-subjects variable was Group, while the within-subjects variable was Posture. The dependent variables were IAc and IS.

Tools

Polar Team2 heart rate monitoring device

The present study employed the Polar Team2 Heart Rate Monitoring System to quantify cardiac IAc through an adapted heartbeat tracking paradigm3. Following a 5-minute seated resting period to ensure cardiovascular stabilization, participants completed timed heartbeat perception trials under controlled laboratory conditions. During experimental trials, initiated by standardized auditory cues, participants verbally reported perceived heartbeat counts across three temporal intervals (25, 35, and 45 s) without physical facilitation (e.g., manual pulse palpation or thoracic contact). Inter-trial recovery intervals of 60 s were systematically implemented to attenuate cumulative fatigue and potential practice effects. Veridical cardiac measurements were synchronously recorded via the Polar Team2 system, a chest-strap heart rate monitoring system validated against electrocardiography in prior studies. IAc indices were computed using the formula: Absolute Percentage Error (APE) was calculated as (|reported heartbeats − recorded heartbeats| / recorded heartbeats) × 100. An individual’s IAc score was calculated as the mean APE across all trials; lower percentages indicate better performance, with 0% reflecting perfect interoceptive sensitivity.

Multidimensional assessment of interoceptive awareness (MAIA)

The Multidimensional Assessment of Interoceptive Awareness (MAIA), developed and revised by Mehling et al.,35,36 is a widely used comprehensive tool for evaluating individuals’ perception of internal physiological signals. In this study, the MAIA was utilized to measure participants’ IS. The scale consists of eight dimensions: Noticing, Not-Distracting, Not-Worrying, Attention Regulation, Emotional Awareness, Self-Regulation, Body Listening, Trusting. The scale comprises 32 statements, each rated on a 0 to 5 Likert scale, where “0” indicates “never” and “5” indicates “always”. Higher scores reflect greater IS. The MAIA has been translated into Chinese and demonstrated good reliability and validity in psychometric studies36.

Experimental procedure

The study protocol was approved by the Ethics Committee for Sports Science Experiments of Beijing Sport University (Approval Number: 2023172 H). Written informed consent was obtained from all participants prior to the commencement of the study.

Baseline Assessment: Following a 5-minute seated rest period in a controlled laboratory, participants completed standardized IAc and IS evaluations.

Intervention Protocol: The three postural conditions were completed across three sessions in a counterbalanced sequence using a Latin-square design. Sessions were separated by 48–72 h to minimize short-term carryover (e.g., transient physiological arousal) while maintaining stable baseline characteristics across conditions. For each participant, sessions were scheduled at the same time of day, and participants were instructed to avoid vigorous exercise, caffeine, and alcohol for 24 h before testing. During each session, participants maintained one of three postures for 10 min: (1) Hunyuan Zhuang Posture: Maintain an upright head posture, slightly tuck the chin, relax the shoulders (creating a subtle hollow at the shoulder joint), and allow the elbows to expand outward and downward. The arms should be in a “hugging” position, with wrists straight, fingers relaxed, hips open, thighs rounded, waist relaxed, pelvis tucked, tailbone slightly dropped, knees slightly bent, aligned with the feet, and not extending beyond the toes. Refer to Fig. 1a for details; (2) Contractive Posture: Stand with legs crossed and arms crossed in front of the torso. Refer to Fig. 1b for details; (3) Neutral Posture: Stand naturally with legs straight and arms relaxed at the sides. Refer to Fig. 1c for details.

Post-Intervention Assessment: Following a 5-minute seated rest period after each intervention, participants repeated the IAc and IS assessments using identical protocols to baseline measurements.

The specific procedure is illustrated in Fig. 2.

Fig. 1
Fig. 1
Full size image

Body Posture Illustrations.

Fig. 2
Fig. 2
Full size image

Flow of participants through screening, counterbalanced sessions, and assessments.

Statistical analysis

All analyses were two-sided with α = 0.05, and data are presented as M ± SD. Baseline differences between experts and novices were tested using independent-samples t tests. Acute effects were examined using separate mixed-design ANOVAs for each outcome with Group (expert vs. novice) as a between-subject factor and Posture (Hunyuan Zhuang vs. contractive vs. neutral) as a within-subject factor. Significant omnibus effects were followed by simple-effects testing and Bonferroni-adjusted pairwise comparisons where appropriate, with effect sizes reported as partial η².

Results and analysis

Differences in IAc

Baseline differences between experts and novices

The M ± SD for the baseline APE scores for each group were as follows: expert group (23.95 ± 11.32) and novice group (33.63 ± 16.69). A t-test indicated a statistically significant difference between the two groups (t = -2.54, p = 0.010), suggesting that the expert group had superior IAc at baseline.

Differences between experts and novices across different postures

As illustrated in Table 2, both expert and novice groups exhibited superior IAc during the Hunyuan Zhuang posture relative to the contractive and neutral postural conditions. Building upon these initial observations, we proceeded to examine how expertise and posture collectively influence interoceptive capacity. Mauchly’s test indicated no violation of sphericity (W = 0.96, χ²(2) = 2.26, p = 0.321), supporting the use of standard univariate ANOVA results. The repeated-measures ANOVA revealed a significant main effect of posture (F(2,112) = 40.57, p < 0.001, η² = 0.42), reflecting substantial differences in IAc across postural conditions. Post-hoc analyses confirmed that all pairwise comparisons between postures were statistically significant: Hunyuan Zhuang versus contractive (MD = -12.00, p < 0.001), Hunyuan Zhuang versus neutral (MD = -17.60, p < 0.001), and contractive versus neutral (MD = -5.50, p = 0.017). Together, these results indicate a clear graded pattern: IAc was highest (APE lowest) during the Hunyuan Zhuang posture, intermediate during the Contractive posture, and lowest during the Neutral posture. In contrast, the group × posture interaction effect was non-significant (F(2,112) = 0.46, p = 0.633, η² = 0.00), indicating that the effect of posture on IAc was consistent across both groups.

The corresponding results are illustrated in Fig. 3a and b.

Table 2 Descriptive Statistics of APE across different groups and postures (M ± SD).
Full size table
Fig. 3
Fig. 3
Full size image

Interoception scores across different groups and postures.

Note

(a) Expert group; (b) Novice group; (c) Expert group; (d) Novice group.

Differences in IS

Baseline differences between experts and novices

The M ± SD for the baseline IS scores for each group were as follows: expert group (3.01 ± 0.24) and novice group (2.80 ± 0.35). A t-test revealed a statistically significant difference between the two groups (t = 2.52, p = 0.015), indicating that the expert group demonstrated superior IS at baseline.

Differences in IS between experts and novices across different postures

As summarized in Table 3, descriptive statistics of MAIA scores indicated elevated IS during the Hunyuan Zhuang posture relative to both contractive and neutral postures across groups. To further evaluate the effects of posture and group on IS, inferential analyses were conducted. Mauchly’s test confirmed sphericity was met (W = 0.98, χ²(2) = 1.09, p = 0.570), supporting the use of parametric repeated-measures ANOVA. A significant main effect of posture was found (F(2,112) = 9.32, p < 0.001, η² = 0.14), reflecting substantial modulation of IS by postural condition. Post-hoc comparisons revealed that IS was significantly higher during the Hunyuan Zhuang posture than during both the contractive (MD = 0.23, p < 0.001) and neutral (MD = 0.25, p < 0.001) postures. No significant difference was detected between contractive and neutral postures (MD = 0.01, p = 0.742). In contrast, the group × posture interaction was not significant (F(2,112) = 2.15, p = 0.123, η² = 0.03), indicating that group differences in IS remained consistent across postures. These results indicate that the Hunyuan Zhuang posture acutely enhances IS, surpassing both alternative postural configurations.

The corresponding results are illustrated in Fig. 3c and d.

Table 3 Descriptive Statistics of IS Across Different Groups and Postures (M ± SD).
Full size table

Experiment 2: effects of long-term hunyuan zhuang intervention on interoception

This study utilized a randomized controlled trial to assess the impact of an 8-week Hunyuan Zhuang intervention on IAc and IS. The findings provide empirical evidence to enhance psychological intervention strategies targeting interoception. This pilot trial was not prospectively registered.

Methods

Participants

A priori power analysis was conducted using G*Power 3.1 for a repeated-measures ANOVA with a within-between interaction. With parameters set to a medium effect size (f = 0.25, η² ≈ 0.06), α = 0.05, and power (1 - β) = 0.80, the analysis indicated a minimum required total sample size of 34 participants. We aimed to recruit approximately 60 participants to account for potential dropouts. Participants were recruited from three Beijing institutions (China University of Geosciences (Beijing), China Conservatory of Music, and Beijing University of Technology) via campus advertisements and course announcements between April and May 2024. Inclusion criteria were: (1) undergraduate students aged 18–25 years; (2) no prior experience in mind-body exercises or martial arts; and (3) self-reported good physical and mental health. Exclusion criteria were: (1) any known cardiovascular, musculoskeletal, or psychological condition that could impede exercise participation; (2) regular use of medication affecting heart rate or attention; (3) a history of competitive athletic training; or (4) any planned absence likely to disrupt the 8-week intervention schedule. Recruitment yielded 69 volunteers. Following eligibility screening, 3 individuals were excluded due to prior practice exposure, and 7 were excluded for anticipated protocol non-adherence. The remaining 59 participants proceeded to randomization. Trial completion revealed 8 discontinuations (attrition rate: 13.6%), primarily due to personal commitments, resulting in 51 analyzable cases (intervention group: 15 M/12F, n = 27; control group: 12 M/12F, n = 24). Full demographic profiles and baseline characteristics are detailed in Table 4. Post-intervention assessments were completed between August and September 2024.

Table 4 Baseline demographic information of participants (M ± SD).
Full size table

Experimental design

A 2 (Group: Hunyuan Zhuang group, Control group) × 2 (Time: Pre-test, Post-test) mixed-design was employed. The between-subject variable was group, and the within-subject variable was time. The dependent variables were IAc and IS. This was an exploratory, parallel-group pilot randomized controlled trial with a 1:1 allocation ratio.

Tools

Polar team2 heart rate monitoring device

As described in Experiment 1, this device was used to measure IAc.

Multidimensional assessment of interoceptive awareness (MAIA)

As described in Experiment 1, this scale was used to assess IS.

Hunyuan zhuang intervention protocol

The 8-week Hunyuan Zhuang intervention was jointly administered by a nationally recognized Chinese martial arts master and a professor from the School of Martial Arts at Beijing Sport University. The protocol followed a standardized, progressively structured training schedule consisting of four sessions per week: three supervised group sessions and one independent home-based session. Supervised group practices were held in a climate-controlled martial arts facility at the China University of Geosciences (Beijing) on Mondays, Wednesdays, and Fridays (19:00–20:00), while participants completed self-guided practice in self-selected environments (e.g., dormitories, outdoor spaces) over the weekend (Saturday or Sunday). Each session was structured into three phases: warm-up, main practice, and cool-down. Detailed instructional procedures are presented in Table 5.

Table 5 Standardized hunyuan zhuang intervention training procedures.
Full size table

Experimental procedure

The study received ethical approval from the Ethics Committee for Sports Science Experiments of Beijing Sport University (Approval Number: 2023172 H). Written informed consent was obtained from all participants prior to any study-related procedures.

Baseline Assessment: Following initial measurements of IAc and IS, participants were stratified by age and gender. Randomization: An independent researcher, not involved in recruitment or assessment, generated the allocation sequence using a computer-based random number generator. A block randomization procedure with randomly permuted block sizes of 4 and 6 was employed to ensure balanced group allocation over time. Allocation concealment was ensured using sequentially numbered, opaque, sealed envelopes (SNOSE). After completing the baseline assessment, the principal investigator opened the next consecutively numbered envelope to reveal the participant’s group assignment (Hunyuan Zhuang intervention group or waitlist control group). To preserve allocation concealment, staff responsible for enrollment did not have access to the random allocation sequence, and group assignment was revealed only after baseline assessment by opening the next sequentially numbered envelope.

Intervention Protocol: The intervention group followed an 8-week standardized Hunyuan Zhuang training protocol (see Table 5). Participants in the control group received no intervention and were instructed to maintain their usual lifestyle. As a usual-lifestyle comparator, control participants were asked not to initiate new mind–body training (e.g., yoga, Tai Chi, qigong, meditation) during the 8-week period.

Post-Intervention Assessment: Post-intervention assessments were conducted 48–72 h after the final training session, replicating baseline protocols under identical environmental conditions.

Blinding: Due to the nature of the intervention, participants and intervention providers were not blinded. Statistical analyses were conducted with group labels coded as A/B until the primary analyses were completed. Protocol changes: No important changes to outcomes, eligibility criteria, intervention procedures, or the analysis plan were made after the trial commenced. Harms: Adverse events were not systematically monitored; any unsolicited reports of exercise-related discomfort or adverse events during the intervention period were recorded. Interim analyses/stopping: No interim analyses were planned or performed, and no stopping guidelines were specified.

The detailed experimental procedure is illustrated in Fig. 4.

Fig. 4
Fig. 4
Full size image

Schematic presentation of the flow of participants through screening, randomization, and the interventions.

Statistical analysis

All analyses were two-sided with α = 0.05, and data are presented as M ± SD. The trial specified two co-primary outcomes, which were analyzed separately using mixed-design ANOVAs with Group (Hunyuan Zhuang vs. control) as a between-subject factor and Time (pre-test vs. post-test) as a within-subject factor. The primary inferential focus was the Group × Time interaction for each co-primary outcome. When interactions were significant, follow-up simple-effects analyses tested within-group pre–post change and between-group differences at each timepoint, with Bonferroni adjustment for multiple comparisons. Effect sizes are reported as partial η², and analyses were conducted on complete cases without imputation for missing post-test data.

Results and analysis

Intervention effects on IAc

Descriptive statistics for APE scores across assessment timepoints are presented in Table 6. The Hunyuan Zhuang group exhibited a decrease in APE from pre- to post-test, whereas the control group showed minimal change over time. A mixed-design ANOVA revealed a significant time × group interaction (F(1,49) = 6.17, p = 0.01, η² = 0.11). Simple effects analysis confirmed that the Hunyuan Zhuang training produced a significant improvement in IAc (F(1,49) = 11.03, p = 0.002), while the control group exhibited no significant change (F(1,49) = 0.10, p = 0.752). These patterns are visually presented in Fig. 5a and b. Significant main effects were also observed for time (F(1,49) = 4.07, p = 0.04, η² = 0.07) and group (F(1,49) = 13.25, p < 0.001, η² = 0.21). Across conditions, post-test APE were significantly lower than baseline values. Moreover, the Hunyuan Zhuang group consistently demonstrated lower APE (higher accuracy) than the control group throughout the study.

Table 6 Descriptive Statistics for APE in the experimental and control groups (M ± SD).
Full size table
Fig. 5
Fig. 5
Full size image

Differences in IAc Between Pre- and Post-Test for the Hunyuan Zhuang and Control Groups. Note (a) Hunyuan Zhuang group; (b) Control group; (c)Hunyuan Zhuang group; (d) Control group.

Intervention effects on IS

Descriptive statistics for IS, as measured by the MAIA, are presented in Table 7. The Hunyuan Zhuang group showed an increase in IS from pre- to post-test, whereas the control group displayed minimal change. A repeated-measures ANOVA revealed a significant time × group interaction (F(1,49) = 8.37, p = 0.006, η² = 0.15). Analysis of simple effects indicated that the Hunyuan Zhuang training resulted in significant enhancement of IS (F(1,49) = 14.10, p < 0.001), while no significant change occurred in the control group (F(1,49) = 0.19, p = 0.665); these results are illustrated in Fig. 5c and d. A significant main effect of time was also observed (F(1,49) = 5.06, p = 0.029, η² = 0.09), reflecting a general increase in IS across timepoints. In contrast, the main effect of group was not significant (F(1,49) = 3.58, p = 0.06, η² = 0.07), indicating that baseline IS levels did not differ substantially between groups. These findings demonstrate that the 8-week Hunyuan Zhuang training program effectively enhanced self-reported IS.

Table 7 Descriptive Statistics of IS for the experimental and control groups (M ± SD).
Full size table

Discussion

This study represents the first systematic investigation of the impact of Hunyuan Zhuang—a foundational training modality in traditional Chinese sports—on interoceptive processing. As a mind-body practice integrating physical postures with cognitive awareness cultivation, the present investigation employed an expert-novice paradigm integrated with randomized controlled design. Results demonstrate that both acute implementation and chronic 8-week Hunyuan Zhuang practice produce significant enhancement in interoceptive capacity. These results suggest that this practice not only induces immediate enhancements in bodily awareness but also facilitates progressive, sustained neurocognitive adaptations through prolonged training. The current findings establish novel theoretical foundations for interoception-targeted interventions while providing empirical validation for the integration of traditional Chinese movement practices into contemporary health paradigms, thereby advancing the scientific globalization of cultural exercise traditions. By demonstrating both expert-novice differences and training-induced changes in IAc and IS, the present work therefore extends prior evidence from Western sports, dance, yoga, mindfulness and clinical interoceptive training programmes into the domain of Chinese traditional standing practices.

Acute mechanisms: optimizing the state of interoceptive inference

The immediate boost in IAc and IS during the Hunyuan Zhuang posture signifies a rapid optimization of interoceptive inference. We posit this acute effect is primarily driven by the enhancement of bottom-up signal precision, orchestrated by its core components.

First, the expansive postural configuration in Hunyuan Zhuang may shift affective and attentional states in a way that increases the salience of internal bodily cues. However, the empirical literature on “expansive vs. contractive” displays indicates more consistent effects on self-reported affect/behavior than on autonomic physiology, and therefore claims about “autonomic tone” should be framed as plausible but not directly measured here30,37. Within predictive processing accounts of interoception, acute improvements can be conceptualized as a temporary increase in the precision-weighting of interoceptive prediction errors—i.e., internal sensory evidence is granted greater influence relative to prior beliefs, enabling more veridical perception9,38. Second, slow, diaphragmatic breathing is a well-established lever on autonomic and psychophysiological regulation, and can improve the signal-to-noise characteristics of cardiorespiratory afferents via vagal pathways. This is consistent with systematic evidence linking slow breathing to enhanced parasympathetic activity and improved emotional regulation31. At the neural level, human intracranial and electrophysiological work shows that breathing phase can entrain/phase-lock activity across distributed cortical-limbic networks, including regions implicated in interoceptive processing and attention, providing a mechanistic route by which breath regulation can facilitate interoceptive monitoring39. Third, sustained, internally directed attention plausibly operates as a top-down attentional set that stabilizes interoceptive focus and reduces distraction by exteroceptive cues. This aligns with mindfulness research showing that training interoceptive attention can alter cortical representations of internal bodily signals32 and with theoretical syntheses emphasizing how contemplative attention reshapes bodily awareness33.

Taken together, we interpret the acute Hunyuan Zhuang effect as a state shift toward higher interoceptive precision and more stable interoceptive attention—an inferential context in which bottom-up bodily evidence is amplified and more consistently sampled. Importantly, because the present study did not directly quantify autonomic markers (e.g., HRV, baroreflex sensitivity) or neural activity, these mechanisms should be treated as testable hypotheses for future multimodal work rather than definitive causal explanations40,41,42.

Long-term adaptation: rewiring the interoceptive generative model

In the present study, both the 8-week Hunyuan Zhuang training and long-term practice by expert participants were associated with significantly greater IAc and IS, indicating that sustained structured practice can enhance the perceptual sensitivity and interpretative awareness of internal bodily states. These findings align with recent meta-analytic evidence showing that mindfulness and related body-focused interventions are associated with improvements in self-reported interoceptive awareness across diverse samples and intervention modalities, suggesting that embodied practice can cultivate interoceptive capacities relevant to broader psychological functioning43.

One plausible mechanism for these lasting improvements involves the optimization of internal generative models that support interoceptive inference. Contemporary theoretical frameworks conceive interoception as hierarchical Bayesian inference, wherein the brain continuously generates predictions about internal states and updates these predictions based on afferent sensory evidence42. Through repeated engagement with structured bodily awareness—such as attending to posture, breathing, and somatic sensations during training—practitioners may provide reliable, high‑precision internal feedback that refines prediction models, leading to reduced prediction error and more accurate bodily representations. This accounts for why long-term embodied practice results in more precise and stable interoceptive processing that is observable even after a single post‑training assessment.

Neurobiological evidence further supports the view that contemplative training can induce plastic changes in networks implicated in interoceptive processing. Functional connectivity studies have shown that mindfulness training strengthens structural and functional integration of the insular cortex—a core hub for interoceptive signal integration—and its connections with broader salience and control networks44. This enhanced insula connectivity may facilitate more efficient integration of internal physiological signals, thereby supporting improved interoceptive awareness and higher IAc and IS scores following sustained practice. Additionally, long-term meditation experience has been associated with structural differences in cortical regions linked with body awareness and sensory integration45, which may further underlie durable enhancements in interoceptive capacity.

Beyond representational improvements, prolonged training may also foster the automatization of interoceptive inference, reducing reliance on effortful attentional control. As generative models become more finely tuned through practice, the process of aligning predictions with incoming signals likely becomes more efficient and less dependent on conscious modulation. This parallels skill acquisition in other domains, where repeated practice transitions cognitive processing from controlled to automatic forms. Evidence from neuroimaging indicates that mindfulness and interoceptive attention engage a distributed network including the insula, anterior cingulate cortex, and prefrontal regions, with patterns of activation modulated by practice experience46. Such neural efficiency suggests that chronic training may reorganize processing dynamics, allowing trained individuals to maintain interoceptive performance with reduced cognitive load and more automatic engagement of relevant neural circuits.

Several limitations of this study should be acknowledged. First, the homogeneous sample consisting exclusively of Chinese university students may limit the generalizability of our findings, highlighting the need for future cross-cultural replications with more diverse demographic characteristics. Second, although the 8-week intervention produced significant effects, the absence of long-term follow-up assessments prevents conclusions regarding the sustainability of these benefits; subsequent studies should therefore incorporate longitudinal designs to examine the persistence of training effects. Third, while we utilized both behavioral and self-report measures, the potential for self-report biases remains; future research would benefit from multimodal approaches integrating neuroimaging, physiological monitoring, and behavioral tasks to enhance methodological rigor. Fourth, the study was not prospectively registered. Future trials should preregister hypotheses, primary outcomes, and analysis plans, and adopt stronger open-science practices (e.g., sharing anonymized data and code) to reduce analytic flexibility and improve reproducibility. Finally, the current study focused exclusively on IAc and IS, without addressing interoceptive awareness. This limitation arises from our use of a heartbeat tracking paradigm, which provides a valid assessment of IAc but is less suited for evaluating metacognitive aspects of interoception compared to discrimination-based protocols incorporating confidence ratings. Future investigations should employ heartbeat discrimination tasks with concurrent confidence assessments to provide a more comprehensive examination of the potential effects of Hunyuan Zhuang practice on metacognitive interoceptive processes.