In the past decades, the study of mindfulness meditation in secular populations has evolved into a popular research topic. What drives people to mindfulness practice in the first place is a deep yearning for inner peace, hiding beneath the indulgence in pleasure and the avoidance of pain. However, this yearning can easily fall into an illusion that inner peace can be gained by estranging thoughts and feelings or alienating self from others. Despite the fact that life circumstances fluctuate, people still hold the belief that they can control future experiences to be replete with joy and free from adversity. Partly due to these misconceptions, psychological problems surge including general distress (e.g., depression, anxiety, stress, worry), negative self-referential processing (e.g., rumination), dysfunctional thinking (e.g., avoidance, suppression), and emotional upheavals (e.g., dysphoria, burnout). Prior research on mindfulness has made big strides in informing us of its multiple facets, correlates, and physical and psychosocial outcomes (e.g., for reviews, see Carpenter et al., 2019; Hölzel et al., 2011; Mattes, 2019; Vago & Silbersweig, 2012; van der Velden et al., 2015).

Adapted to Western societies, the formulation of secular mindfulness practice draws on and also deviates from Buddhist traditions. Mindfulness is described as a quality of awareness that is open, receptive to what is unfolding without being swept away by judgment and ingrained reactivity (Baer, 2003; Brown & Ryan, 2003; Segal et al., 2002). In Buddhist discourse, mindfulness refers to directly observing and experiencing what is spontaneously present without being distracted by craving, aversion, or fixation on mental events (Bodhi, 2000; Olendzki, 2011). This bare awareness has a temporal quality of immediacy and a spacious quality of the expansiveness of mind. The features common to mindfulness-based trainings include focusing attention on an object in the present moment, monitoring awareness of attention without judgment, bringing attention back from the distractor, and refocusing on the target without reactivity (Kabat-Zinn, 2003). Contemporary secular mindfulness is practiced as a means to reduce stress, treat chronic pain among patients, prevent depression relapse among clinical populations, and increase well-being for the general public.

Evidence is accumulating that secular mindfulness practice generates subjectively felt and physiologically observed health benefits. Empirical studies found that that trait mindfulness was positively related to the sense of flourishing in athletes (e.g., Zhang et al., 2021), savoring and meaning of life in cancer patients (e.g., Garland et al., 2017), positive coping strategies (e.g., Dillard & Meier, 2021; Kock et al., 2021), emotional well-being (e.g., Mahlo & Windsor, 2021), and life satisfaction (e.g., Waterschoot et al., 2021) at times of adversity. Moreover, mindfulness mitigated depressive symptoms through a decrease in rumination and negativity bias (e.g., Paul et al., 2013), and predicted higher well-being via a reduction in self-stigma (e.g., Wang et al., 2023). An experience-sampling study (Shoham et al., 2017) found that over the course of a six-session training, community samples became more mindful of inner experiences, which in turn tempered emotional arousal and elevated positive emotional states. Increasing the frequency of daily mindfulness practices had stress-buffering effects (e.g., Manigault et al., 2021). Conversely, mindfulness was negatively associated with psychological problems such as depressive symptoms, anxiety symptoms, and distress (e.g., Gecht et al., 2014; Ma & Siu, 2020; Pearson et al., 2015). In a similar vein, neuroscientific studies showed that mindfulness was associated with improved self-regulation, as indicated by more activities in prefrontal cortex implicated in executive control and less activation in amygdala involved in emotional reaction (e.g., Creswell et al., 2007). Also, mindfulness facilitated the transition from self-referential processing to momentary experiencing, as indicated by lower brain activities in areas (e.g., medial prefrontal cortex) related to conceptual processing and elevated activities in areas (e.g., insula, inferior parietal lobule) related to experiencing (e.g., Berkovich-Ohana & Glicksohn, 2014; Farb et al., 2007). Together, these findings are united in highlighting that mindfulness can bolster well-being and reduce psychological problems.

Taking a step further, researchers explore cognitive, emotional, and behavioral mechanisms central to the effects of mindfulness. The proposed mechanisms include intention, attention regulation, non-attachment, decentering, emotion regulation, and compassion (e.g., Grabovac et al., 2011; Ho et al., 2022; Hölzel et al., 2011; Shapiro et al., 2006; Vago & Silbersweig, 2012) and the combination of acceptance and attention monitoring (e.g., Lindsay & Creswell, 2019). Among these mechanisms, decentering merits further investigation. It was theorized that cultivating mindfulness prepares one to shift perspectives and approaches towards mental activities, namely, decentering (Bishop et al., 2004; Shapiro et al., 2006). This view was in line with prior findings that mindfulness and decentering were distinct and interrelated processes (e.g., Gecht et al., 2014; Hayes-Skelton & Graham, 2013). Another plausible interpretation was that mindful awareness and attentiveness were preconditions for decentering. It is because attention focus and stability, which is fostered through mindfulness practice, stills and calms the mind. This mental balance sharpens awareness of inner states, making it possible for people to inhibit from perpetuating negative thoughts (e.g., Bernstein et al., 2015; Grabovac et al., 2011). Empirical studies supporting this notion found that mindfulness was closely associated with decentering in daily life (e.g., Biehler & Naragon-Gainey, 2022) as well as during pandemic (e.g., Kock et al., 2021; Waterschoot et al., 2021). Four facets of mindfulness (describing, non-judgment, non-reactivity, acting with awareness) predicted higher levels of decentering, which in turn predicted distress tolerance, depressive symptoms, anxiety symptoms, and stress (e.g., Brown et al., 2015). To fully understand what decentering means, it is necessary to trace its roots back to Buddhist traditions, which helps us to see the points of convergence and divergence between traditional Buddhist teachings and contemporary Western psychological perspectives.

One salient difference is that Buddhist traditions take the dialectical approach to unveil what are the hindrances to effective decentering, while Western psychology endeavors to examine how decentering works. Within Buddhist frameworks, a major blockage of decentering rests on the lack of understanding of the true nature of human existence. The concept of decentering encapsulates the essence of the three marks of existence: impermanence, suffering, and no-self (Gyatso, 2000). Firstly, all conditioned phenomena are impermanent. There are causes, enabling conditions and corresponding circumstances in which things originate and vanish. Both external matter and internal states are impermanent in the sense of being subject to a constant state of flux. To see things as they really are, we need to decenter from the illusion of permanence, including our self-related conceptions. Secondly, all contaminated phenomena are suffering. Phenomena under the control of our negative emotions and thoughts become contaminated and produce suffering for ourselves and others. Our mind is prone to afflictive emotions (craving, aversion, ignorance, and delusion), leading to discontentment with our experiences and distorted perception. Humans tend to cling to favorable objects and exaggerate their desirable qualities, which results in misattribution of suffering to the absence of desirable objects rather than to the content of one’s mind. Realizing that causes of suffering reside within our undisciplined mind is a prerequisite for us to decenter from and uproot them. Thirdly, all phenomena are empty of self-existence (no-self). Emptiness, here, means that things do not come into existence by means of their own power. Decentering from the causes of suffering requires us to penetrate into the illusion that things are intrinsically existent, including one’s self.

The Buddhist doctrine of no-self illuminates a key hindrance to decentering. No-self is not to equate the existential status of self as simply being nothing, but to decenter from self-delusion (Bodhi, 2000; Thich Naht Hanh, 1998). By examining the various parts of the self, Buddhist traditions (e.g., Madhyamaka school in Mahāyāna Buddhism, Dzogchen in Tibetan Buddhism) understand self in terms of five skandhas, aggregates of elements (body, sensations, perceptions, mental formations, and consciousness), which make up individual physical and mental existence (Gyatso, 2000). Given that the five aggregates are constantly changing, self is an ongoing process of transformation.

Based on this notion, Mahāyāna Buddhism conceptualizes self in terms of conventional truth and ultimate truth (Dunne, 2011). In conventional truth, there exists a functioning self who manages internal conflicts and confusion, performs duties, seeks feedback, weighs options, and adjusts actions in this world. However, in ultimate truth, self is devoid of intrinsic existence. It is in this context that we are awakened to the fact that the identities and self-concepts are cognitive representations, so they are empty in nature, akin to images and events in dreams (Ryan & Rigby, 2015). With this realization comes the greater capacity to release the defensive grasping to beliefs and self-absorption.

It is in this spirit that Buddhism discusses decentering in terms of stepping back from mental events and stepping outside of self-centered frame of reference. In comparison, Western psychology conceptualizes decentering mainly as disidentification and non-reactivity to mental events, but what remains unaddressed is the centrality of self in the process of creating mental content (Dahl et al., 2015). From the Buddhist philosophical standpoint, what lies at the root of aversion, craving, and ignorance is not simply the lack of attention, but self-delusion which involves deliberate misinterpretation of self and the world (Dunne, 2011; Gethin, 2015). What presents itself to us is different from what it really is because we re-present it through the filters of our conditioned beliefs, preferences, and preconceptions. Thus, self-delusion reflects our attempts to maintain self-identities, which satisfies the need for solidarity and stability. As an antidote, decentering in Buddhist frameworks operates at two dimensions: disidentifying from inner states and disengaging from the focal point of the self (Dambrun & Ricard, 2011).

The Western conceptualization of decentering depicts the first dimension, whereas the second dimension awaits further research. For example, the influential model of the metacognitive processes model of decentering (Bernstein et al., 2015) defines decentering as a mental phenomenon which comprises three interrelated mental processes: meta-awareness, disidentification from inner experiences, and reduced reactivity to thoughts. Specifically, meta-awareness refers to being aware of the cognitive process (sensing, perceiving, thinking, feeling) that generates internal states (e.g., sensations, perceptions, thoughts, feelings). Disidentification from inner experience refers to experiencing inner states as integral parts of the self. That is, one cannot look upon thoughts and feelings as self, or look upon self as thoughts or feelings (Hölzel et al., 2011). Finally, reduced reactivity to thoughts refers to the weakened power of thoughts on other mental processes (e.g., feeling). Being able to separate subjective interpretations of experiences from reality can help individuals to accept and tolerate undesirable situations, and break the cycle of repetitive negative thoughts and excessive self-absorbed attention (Bernstein et al., 2019; Bishop et al., 2004). However, the impact of a centering “I” on constructing, maintaining, and defending mental activities is not explicated in the model.

Despite the discrepancy between Buddhist traditions and Western theoretical account of decentering, there is growing evidence showing the impact of decentering on psychological health. Applying ecological momentary assessment, a recent study collected daily surveys from community members (e.g., Sgherza et al., 2022). The results showed that the current levels of decentering predicted well-being at a later time point, and the current levels of well-being predicted subsequent decentering, suggesting that decentering and well-being mutually strengthened each other (e.g., Sgherza et al., 2022). In the reverse line, decentering was negatively related to depressive symptoms (e.g., Linares et al., 2016) and math anxiety among undergraduate students (e.g., David et al., 2022). A closer look at the facets of decentering found that changes in disidentification were associated with decreases in depressive and anxiety symptoms (e.g., Bieling et al., 2012), stress (e.g., Hoover et al., 2022), worry (e.g., Kock et al., 2021), rumination (e.g., Fresco et al., 2007), and suppression (e.g., Fuochi & Voci, 2020).

Moreover, it has been proposed that decentering is a mechanism of change linking mindfulness to enhanced subjective well-being and reduced psychological symptoms (Grabovac et al., 2011; Shapiro et al., 2006). Evidence supporting this proposal came from both correlational and intervention studies (e.g., Bieling et al., 2012; Hoge et al., 2015; Pearson et al., 2015). To illustrate, path analysis showed that decentering linked mindfulness to positive affect among undergraduate students (e.g., Ben Salem & Karlin, 2023). Among individuals with no meditation experience and with varying levels of meditation experience, decentering linked the association between meditation practice and value-based behaviors (e.g., Franquesa et al., 2017). Parallel results were reported in intervention studies among healthy and clinical samples. For instance, in a healthy university-community sample, a brief four-session mindfulness training improved trait and state mindfulness, and this improvement had a robust correlation with reduced experiential avoidance and elevated decentering (e.g., Tanay et al., 2012). Among patients seeking treatment for chronic pain, decentering from negative thoughts made the occurrence of pain less distressing and more tolerable, and improved the quality of emotional and psychosocial functioning in daily life (e.g., McCracken et al., 2013). Likewise, a brief 15-min mindfulness induction focusing on breathing had a positive effect on decentering, which in turn, reduced ratings in pain, desire for pain medication, and anxiety among orthopedic surgery patients (e.g., Hanley et al., 2021). In depressed people, a brief induction of experiential self-focus also boosted the level of decentering, compared with a ruminative self-focus (e.g., Lo et al., 2014). Notably, decentering was found to be a shared mechanism across mindfulness-based stress reduction and mindfulness-based cognitive therapies (e.g., Hayes-Skelton & Graham, 2013; Hayes-Skelton & Lee, 2020; Hayes-Skelton et al., 2015). In spite of these promising results, contradictory results were evident in the literature. Studies showed that decentering did not have mediating effects on the relationship of mindfulness with psychological distress (e.g., Bhambhani & Cabral, 2016; Carmody et al., 2009). Given these conflicting findings, it is necessary to further explore and ascertain the role of decentering in mindfulness.

Despite mounting empirical studies, there is still a lack of statistical synthesis to evaluate the strength and consistency of the effect of decentering in linking the association between mindfulness and psychological problems. The aim of this study was thus twofold. The first aim was to investigate the indirect effect of decentering in the relationship between mindfulness and various psychological problems. The second aim was to probe whether this relationship was moderated by age and clinical status.

The current meta-analysis added to prior meta-analytic studies in several ways. Firstly, this study focused on decentering, so as to present a clear picture of the direction and magnitude of its effect. Prior meta-analyses shed lights on a wide array of mediating variables linking mindfulness with psychological functioning. For example, rumination mediated the effects of mindfulness-based interventions on mental health outcomes, such as depressive and anxiety symptoms, stress, and negative affect (e.g., Gu et al., 2015). Additionally, mindful attention, decentering, and acceptance were identified as mediators of the benefits of acceptance and mindfulness-based therapies on depression and anxiety symptoms (Johannsen et al., 2022). The overall mediating effect was significant, and of a small to medium magnitude (r = 0.14, p < 0.001). Noteworthy, the study of Johannsen et al. (2022) tested the mediating effect when mindful attention, decentering, and acceptance were combined, which precluded the examination of the unique mediating effect of decentering. Regarding the time frame, Johannsen et al. (2022) included studies published between 2010 and 2021, whereas the current meta-analysis included studies published until May 2023 to provide a comprehensive review. Furthermore, this meta-analysis probed different individual characteristics (i.e., age, clinical status) as moderators, extending the study of Johannsen et al. (2022) which tested moderators of therapy type, treatment format, problem type, control group, and final assessment point. The findings could advance our understanding of individual responsiveness to treatment, which can be leveraged to tailor the components of interventions for targeted populations. Lastly, the current study applied the new technique of meta-analytic structural equation modeling (MASEM; Jak et al., 2021) because it enabled us to draw conclusions about the potential mediating effect of decentering based on a large pool of studies, in which most studies did not directly test the mediating effect or did not examine all the relevant relations.

Method

Search Strategy

To locate relevant studies, four electronic databases were consulted: PubMed, PsycINFO, Scopus, and ProQuest. The literature search was conducted on May 26–28, 2023, for peer-reviewed journal articles. It was updated on April 25–27, 2024, for doctoral dissertations and the date range was set from onset to May 2023. The search query is provided in Online Resource 1.

Based on the inclusion and exclusion criteria (Table 1), the screening process was performed in three stages. The first stage filtered articles based on title and abstract. Next, the retained articles were screened with respect to whether the association between mindfulness and decentering or the association between decentering and psychological problems were examined. In the last stage, retained articles were scrutinized in full length and special attention was given to data availability for effect size calculation. Regarding the case of missing data in an eligible study, the corresponding author was contacted through email to maximize inclusion rate. Figure 1 depicts the literature search process following the PRISMA guideline (Page et al., 2021).

Table 1 Overview of inclusion and exclusion criteria
Full size table
Fig. 1
figure 1

PRISMA flow diagram of the literature search process

Full size image

Data Extraction

Firstly, Pearson’s correlation coefficient (r) was used as a measure of effect size. Noteworthy, the correlation coefficients were extracted from survey studies that did not include treatment and from mindfulness-based intervention studies that examined bivariate correlations among study variables at baseline.

When a study reported a composite score for the mindfulness measure and also scores for each facet of mindfulness, only the composite score was extracted to keep it consistent across studies. If the composite score is absent, individual scores for each mindfulness facet were averaged. Some studies administered surveys or treatment on different samples, such as multiple student groups, mediators and non-meditators, clinical samples, and non-clinical samples. In these cases, these samples were considered as independent samples and available correlation coefficients were extracted separately.

Meta‑analytic Procedure

Effect Size Calculations

To address the research questions, two types of statistical analyses were performed in the software environment R (R version 4.2.3; R Core Team, 2023). Firstly, separate multi-level meta-analyses with random effects were performed to examine the relations of primary interest (Assink & Wibbelink, 2016). Specifically, this approach estimates variance at three levels: (1) sampling error of a study, (2) variance within a study, and (3) variance between studies. In the current meta-analysis, an independent sample was used as the unit of analysis at the second level because some studies recruited different samples to complete surveys or treatment (e.g., Cladder-Micus et al., 2019; Feliu-Soler et al., 2016; Naragon-Gainey & DeMarree, 2017). The overall effect size for each relation was computed using metafor package (Viechtbauer, 2010), and checked for publication bias and outliers. Since the majority of the included studies examined multiple psychological outcomes, the multi-level modeling approach allows us to control for effect size dependency in the same sample. To this end, robust variance estimation was conducted (Hedges et al., 2010) for the multi-level meta-analyses, using the Robumeta package (Fisher et al., 2017).

Secondly, MASEM (Jak et al., 2021) was applied to examine whether decentering mediated the relationship of mindfulness with psychological problems. For the current study, a random-effects MASEM is an appropriate technique, assuming that each study has its own population correlation matrix (Jak et al., 2021). The structural equation model was evaluated based on the averaged correlations between study variables across studies. The mediation model evaluated the significance and strength of the indirect effect of mindfulness on psychological problems through decentering. The advantage of MASEM is that it enables us to test the mediation model by integrating data from studies that examined part of the relations of interest. For example, some studies reported the correlation between mindfulness and decentering, but not between decentering and psychological outcomes. The R script for MASEM is provided in Online Resource 2.

Moderator Analyses

MASEM can also test whether correlation coefficients vary with study-level moderator variables, which explains between-study heterogeneity (Jak et al., 2021). The moderator analyses of categorical variables were conducted separately in meta-analyses, in which each moderator operated as an independent variable predicting the effect size of each relation. Specifically, the moderating effect of age or clinical status was included in meta-analyses on the direct effect of mindfulness on decentering (Path a), of decentering on psychological problems (Path b), and of mindfulness on psychological problems (Path c′). The parameter estimates applied the metaSEM package (Cheung, 2015) and full-information maximum likelihood estimation based on the OpenMx package (Boker et al., 2011). Concerning the significance of a parameter estimate, if the 95% confidence interval does not include zero, this parameter is considered as significant.

This study probed two individual characteristics as moderators, namely, age (emerging adults vs. older adults) and clinical status (clinical samples vs. non-clinical samples). According to the coding scheme (Table 2), the included studies were coded by two coders independently: one author who coded all the studies and one trained graduate assistant who coded 25% of the studies. The inter-rater reliability was 0.91, as calculated by Cohen’s κ. Discrepancies were later resolved through discussion. The rationale for probing these two moderators was explained below.

Table 2 Coding of moderators
Full size table

Presumably, decentering is a skill that accrues through navigating different domains of life, such as intimate relations, professional identities, and financial basis (Bernstein et al., 2015). Studies provided evidence that well-being for middle adulthood was closely associated with the ability to disengage from unattainable goals and engage with alternative pursuits (e.g., Feliu-Soler et al., 2016). In particular, decentering was positively related to flexible goal adjustment, and the relations grew stronger in middle-aged adults than in young adults (e.g., Mahlo & Windsor, 2021). Likewise, studies found a positive association between observing and affective symptoms and lower levels of non-judging and non-react to inner states among young student samples, which might due to the insufficient development of the capacity to withhold judgment and reaction (e.g., Curtiss et al., 2017; Eisenlohr-Moul et al., 2012). In comparison, older adults exhibited higher levels of non-judging and non-react (e.g., Frank et al., 2015; Prakash et al., 2015). This pattern was more salient in middle adulthood because it is a stage of life when people need to juggle professional, familial, and financial responsibilities and to navigate gains (e.g., growth in intellectual and emotional maturity) and losses (e.g., declines in physical agility and strength) in personal life (e.g., Carstensen, 2006; Lachman et al., 2015). Thus, it was assumed that the effect sizes might be greater in studies with older adults than in studies with emerging adults.

With respect to levels of psychological problems and vulnerability to emotional cues, clinical samples differ from samples of community residents and undergraduate students. Clinical samples are susceptible to cyclical thinking, which exacerbates psychological symptoms such as existential anxiety about one’s flaws and inadequacies (Watkins, 2008). Moreover, rather than cognitive deficits, depressive people are conditioned to use suboptimal strategies such as suppression and rumination, and define themselves in a rigid, defensive manner (e.g., Aldao et al., 2010). Available evidence showed that decentering skills mitigated the risk for excessive negative affect and deficient positive affect among clinical samples with disorders and subclinical samples with symptoms (e.g., Naragon-Gainey & DeMarree, 2017). For healthy samples (e.g., people with moderate levels of positive and negative affect), other strategies such as cognitive reappraisal might be more effective (e.g., Naragon-Gainey & DeMarree, 2017). Thus, it was assumed that stronger effects might be manifested in studies with clinical samples, in contrast to studies with non-clinical samples.

Heterogeneity and Sensitivity Analysis

To evaluate systematic variation of effect sizes across studies, the current study used Q and I2 statistics of the correlation coefficients (Higgins & Thompson, 2002). The Q statistic represents the usual test statistic, with a p-value smaller than 0.05 indicating significant between-study variation. The I2 statistic represents the percentage of the variation that is due to true differences in effect sizes as opposed to within-study sampling error (Borenstein et al., 2010). The recommendation is that I2 values of 25%, 50%, and 75% denote low, moderate, and high levels of heterogeneity, respectively (Higgins & Thompson, 2002). Together, significant Q test results and I2 values higher than 25% with wide confidence intervals point to the necessity of moderator analyses.

To examine selective bias, the current study adopted the PET-PEESE approach using the PETPEESE function in R (Carter et al., 2019). The precision-effect test (PET; Stanley & Doucouliagos, 2014) regresses the effect size estimate on its precision (i.e., standard error), weighted in inverse proportion to the variance (i.e., the square of standard error). Different from the linear regression model in PET, a related approach is precision effect estimate with standard error (PEESE; Stanley & Doucouliagos, 2014) which fits a quadratic relationship of the estimated effect size and the standard error to the data, also weighted by the inverse of the variance. In the approach of PET-PEESE, if the PET estimate is statistically significant in a one-tailed t-test for bias (α = 0.05), it means that the true underlying effect is different from zero. With a non-zero overall effect, the PEESE estimate is preferred to the PET estimate (Hagger et al., 2018).

To detect influential outliers, this study took two approaches: the standardized residual and Cook’s distance, and the leave-one-out approach (Borenstein et al., 2009). When the standardized residual exceeds 3.0 in absolute value and Cook’s distance is greater than 4 / (n – k − 1), the study is deemed as an outlier. The formula for calculating Cook’s distance is 4 / (n – k − 1), with n denoting the total effect sizes and k the number of predictors (Fox, 1991). In addition, when removing a study from the meta-analysis dramatically changes the overall effect size or heterogeneity, this study is considered as an influential outlier and omitted from further statistical analyses.

Results

Study Selection and Characteristics

This meta-analysis included 110 effect sizes extracted from 57 independent studies, including peer-reviewed journal articles and doctoral dissertations. The total sample size was 18,515, ranging from 19 to 1891 participants. The current pool of studies comprised two categories: (1) correlational studies (n = 37) that examined the association between mindfulness and decentering, or between decentering and psychological problems, or both; and (2) intervention studies (n = 20) that measured bivariate correlations between study variables at baseline. The interventions had three types: mindfulness-based stress reduction, mindfulness-based cognitive therapy, and others (i.e., mindfulness meditation, mindful breathing, mindfulness-integrated cognitive behavior therapy). Table 3 presents an overview of the characteristics of the included studies.

Table 3 Overview of the characteristics of the included studies
Full size table

The widely used measures of mindfulness include the Five Facet Mindfulness Questionnaire (FFMQ; Baer et al., 2006) which taps into five facets of dispositional mindfulness: observing, describing, acting with awareness, non-judgment, and non-reactivity and the Mindful Attention Awareness Scale (MAAS; Brown & Ryan, 2003) which assesses dispositional mindfulness in terms of awareness of and attention to the present moment. The two commonly used measures of decentering are the Experiences Questionnaire-Decentering Subscale (EQ; Fresco et al., 2007) and Toronto Mindfulness Scale-Decentering Subscale (TMS; Lau et al., 2006). The measures of psychological problems encompass Beck Depression Inventory-II (BDI-II; Beck et al., 1996); Depression, Anxiety, and Stress Scale (DASS-21; Lovibond & Lovibond, 1995); Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001); Generalized Anxiety Disorder-7 (GAD-7; Spitzer et al., 2006); Center for Epidemiologic Studies Depression scale (CES-D; Radloff, 1977); Penn State Worry Questionnaire (PSWQ; Meyer et al. 1990); Perceived Stress Scale (PSS; Cohen et al. 1983); Response Style Questionnaire (RSQ; Nolen-Hoeksema, 1991); and the Positive and Negative Affect Scale (PANAS; Watson et al., 1988).

Multi-Level Modeling

To compute the overall effect size, separate meta-analyses were conducted for each relation (Table 4). For the relation between mindfulness and decentering (Path a), the overall effect size was r = 0.44 (95% CI[0.36, 0.52]), suggesting a positive, medium-sized effect (see Fig. 2 for the forest plot). That is, higher levels of mindfulness were associated with higher levels of decentering, which accounted for 19% of the variance in decentering. Given that the overall effect size in the PET test was significantly different from zero, the PEESE regression test was taken to examine publication bias. The result suggested that publication bias seemed to be present in the effects (see Fig. 3 for the funnel plot). The PEESE estimate (r = 0.41, 95% CI[0.35, 0.48]) after adjusting for bias was slightly lower than the overall estimate of effect size from the meta-analysis, and the effect was still in the positive direction and medium-sized. Thus, the conclusion was not changed with respect to the overall effect.

Table 4 Results of random-effects univariate meta-analyses
Full size table
Fig. 2
figure 2

Forest plot of the effect sizes with 95% confidence interval for Path a

Full size image
Fig. 3
figure 3

Funnel plot for Path a

Full size image

Regarding the relation between decentering and psychological problems (Path b), the overall effect size was r =  − 0.36 (95% CI[− 0.40, − 0.32]), suggesting a negative, small effect (see Fig. 4 for the forest plot). That is, higher levels of decentering were associated with lower levels of psychological problems, which accounted for 13% of the variance in psychological problems. One study (Ben Uriel-Maoz et al., 2023) was identified as an influential outlier, and removed from analysis. There was significant heterogeneity in the effect sizes (p < 0.0001), and the systematic variation between studies were moderate (I2between = 64.73%). The PET-PEESE test suggested that publication bias was unlikely in the effects (see Fig. 5 for the funnel plot). The bias-corrected PEESE estimate (r =  − 0.36, 95% CI[− 0.40, − 0.31]) was very close to the meta-analytic estimate.

Fig. 4
figure 4

Forest plot of the effect sizes with 95% confidence interval for Path b

Full size image
Fig. 5
figure 5

Funnel plot for Path b

Full size image

As to the relation between mindfulness and psychological problems (Path c′), the overall effect size was r =  − 0.37 (95% CI[− 0.42, − 0.31]), suggesting a negative, small effect (see Fig. 6 for the forest plot). That is, greater levels of decentering were associated with lower levels of psychological problems, which explained 14% of the variance in psychological problems. One influential outlier (Francis et al., 2022) was detected and omitted from analysis. There was also significant heterogeneity in the effect sizes (p < 0.0001), and the between-study variability was low (I2between = 24.75%). The results of PET-PEESE test indicated that publication bias seemed to be present in the effects (see Fig. 7 for the funnel plot). However, the conclusion regarding the overall effect remained unchanged because the difference between the adjusted PEESE estimate (r =  − 0.35, 95% CI[− 0.41, − 0.30]) and the meta-analytic estimate was marginal, and the effect size was also in the negative direction and of small magnitude. Together, publication bias was not a serious concern in the current results.

Fig. 6
figure 6

Forest plot of effect sizes with 95% confidence interval for Path c′

Full size image
Fig. 7
figure 7

Funnel plot for Path c′

Full size image

Meta-analytic Structural Equation Modeling

The proposed mediation model was fit to the pooled correlation matrix (see Fig. 8 for the mediation model). All parameters output from the MASEM analyses were in standardized metric with 95% likelihood-based confidence intervals. With respect to model fit, the goodness-of-fit statistics showed a saturated model, so the parameter estimates were more informative (Jak et al., 2021). In the current sample, all direct and indirect effects in the model were significantly different from zero (Table 5). Concerning the interpretation of direct effects, coefficients below 0.10 are considered as small, between 0.10 and 0.25 as small-to-medium, between 0.25 and 0.40 as medium-to-large, and above 0.40 as large (Lipsey & Wilson, 2001). In the current sample, mindfulness had a positive, large direct effect on decentering ( β = 0.42, 95% CI[0.38, 0.47]), and a negative, small-to-medium direct effect on psychological problems (β =  − 0.25, 95% CI[− 0.31, − 0.20]). Decentering also had a negative, small-to-medium direct effect on psychological problems (β =  − 0.24, 95% CI[− 0.29, − 0.19]).

Fig. 8
figure 8

The final model

Full size image
Table 5 Standardized parameter estimates for the mediation model
Full size table

Regarding the indirect effect, the effect of mindfulness on psychological problems was mediated by means of decentering, albeit a small magnitude (β =  − 0.10, 95% CI[− 0.13, − 0.08]). Indirect effects equal the product of direct effects, so they are generally of a small size. After including decentering as a mediator, the direct effect of mindfulness on psychological problems remained statistically significant. Thus, the model represented a partial mediating effect through decentering rather than full mediation. The total effect of mindfulness (− 0.35) was calculated by summing the indirect and direct effects, meaning that one unit increase in mindfulness corresponded to a 0.35 standard deviation decrease in psychological problems. Based on the proportion mediation statistic (PM; Ditlevsen et al., 2005), the mediating effect via decentering explained a small proportion of the total effect of mindfulness (PM = 0.29). In other words, 29% of the total effect of mindfulness on psychological problems was accounted for by the indirect effect of decentering.

Moderator Analyses

The current study tested two categorical moderators: age and clinical status. Table 6 presents the parameter estimates from the moderation models at α = 0.05. To compare correlation coefficients between the studies with emerging adults and the studies with older adults, moderator analyses were performed with age as the moderator. The omnibus test of moderating effects showed a non-significant overall effect (χ2(3) = 5.75, p = 0.12), suggesting that age had no significant moderating effects on any of three direct effects.

Table 6 Results of the moderator analyses
Full size table

To compare the correlation coefficient between studies with clinical samples and studies with non-clinical samples, moderator analyses were performed with clinical status as the moderator. The results revealed significant differences in model effects (χ2(3) = 10.59, p = 0.01), suggesting that at least one of the three direct effects was moderated by clinical status. Close inspection of individual parameters found that clinical status only significantly moderated the direct effect of mindfulness on decentering. The estimation of the direct effect of mindfulness on decentering was weaker in studies that involved non-clinical samples, as compared to studies that involved clinical samples (\(\widehat{\beta }\)= − 0.23, 95% CI[− 0.38, − 0.08]). It seemed that mindfulness was a stronger indicator of decentering among participants who were diagnosed with psychological disorders or who exhibited psychological symptoms. However, there were only six effect sizes from studies that recruited clinical samples whereas there were 50 effect sizes from studies that recruited non-clinical samples. The unbalanced number of effect sizes in subgroups might contribute to the moderating effect, which precludes us from drawing a firm conclusion. Taken together, age and clinical status were not significant moderators in the current pool of studies.

Discussion

This meta-analysis applied the statistical technique (MASEM) to investigate the relationship between mindfulness, decentering, and psychological problems. Additionally, moderator analyses examined whether this association varied depending on individual characteristics of age and clinical status. The results suggested that decentering might have a significant, small indirect effect in the association between mindfulness and psychological problems. Regarding age, the effects were consistent across emerging adults and older adults. Regarding clinical status, stronger effects of mindfulness on decentering were identified in clinical samples, as opposed to non-clinical samples. However, given the disproportionately small number of studies with clinical samples in subgroups, the result was not practically significant. Thus, age and clinical status were not significant moderators in the current pool of studies.

The current results provided preliminary support for the claim that decentering is a pathway through which mindfulness confers benefits on psychological outcomes (e.g., Bernstein et al., 2015; Bishop et al., 2004; Shapiro et al., 2006; Vago & Silbersweig, 2012). The findings also corroborated a recent meta-analysis (Johannsen et al., 2022) which showed that decentering together with mindful attention and acceptance had a small-to-medium meditating effect in the treatment of depression and anxiety (r = 0.14). Rather than combining variables, the current meta-analysis parsed out the mediating effect of decentering on a broad spectrum of psychological problems and found a significant, small effect size (β =  − 0.10). Moreover, the study of Johannsen et al. (2022) aggregated data from difference scores between treatment and control groups in mindfulness-based interventions whereas the current study aggregated data from correlation coefficients. Despite these differences, the convergent findings from both meta-analytic studies confirm the role of decentering in the effects of mindfulness on psychological problems.

Statistically, the results of this study showed that decentering had a significant, small effect size. Empirically, decentering might be one of the factors that are positively correlated with mindfulness and negatively correlated with psychological problems. The pattern of correlations can also be produced by other factors representing different psychological processes. For example, decentering were related to other Buddhist-derived concepts such as compassion (e.g., King, 2023) and non-attachment (e.g., Mak et al., 2023), and acceptance (e.g., Liu et al., 2023). Specifically, decentering and acceptance mediated the effects of MBSR on distress tolerance and resilience (e.g., Ma & Siu, 2020; Nila et al., 2016). Improved emotion regulation mediated the effect of decentering on recovery from depression in MBCT (e.g., Bieling et al., 2012), and value-driven behavior mediated the association between decentering and well-being in response to setbacks and stressful conditions (e.g., Rogge et al., 2024). The central idea gleaned from these studies was that besides decentering, other cognitive and motivational variables might as well transmit the effects of mindfulness to psychological outcomes. Future studies can examine these variables as parallel or serial mediators.

The current findings have practical implications. Firstly, mindfulness-based programs should incorporate decentering as an essential element in their training. Decentering provides a safe base for us to relate to inner processes, and build resources to deepen that relationship. As in a warm, open, curious relationship, we are willing to move deeper into our inner experiences, and explore possibilities of experiencing our language, thoughts, and feelings without the fear of being governed by them.

Given that decentering might not be a natural response, it requires training and repeated practice for people to change the way they experience reality across diverse contexts (e.g., Kiken et al., 2015). Empirical evidence supporting this view showed that meditators with 10-year experience demonstrated higher levels of mindfulness and decentering, compared to non-meditators (e.g., Isbel & Mahar, 2015). The training of decentering anchors at the step-back ability that enables practitioners to deconstruct various aspects of experiences (perceptual, cognitive, emotional), and distinguish between what appears and what is cognitive representation (Levit-Binnun et al., 2021). Over time, the repeated practice can promote cognitive flexibility and emotion regulation (e.g., Feldman et al., 2007; Lau et al., 2006; Lo et al., 2014), which pointed to the need of training decentering alongside other skills to bring about changes in everyday life (e.g., Sgherza et al., 2022).

However, one caveat is that practitioners fall back into their habitual groove of thinking and acting after meditation sessions. It is necessary to provide the fertile soil for decentering skills to take root in individual life style. To achieve this, foundational philosophical principles should be explicitly taught before or throughout the course of meditation training, so that practitioners understand how the mind works and why the causes of suffering lie within (Crane, 2017; Van Gordon et al., 2015). A deep understanding of the three marks of existence (i.e., impermanence, suffering, no-self) sets the context in which practitioners discern and voluntarily break the cycle of unwholesome thoughts and emotions, which has rippling effects for them to change routines in daily life (Dunne, 2011; Greenberg & Mitra, 2015).

Conceivably, we need to make a distinction between therapeutic and spiritual goals. The therapeutic endeavor is to help individuals relieve symptoms, ease tension, and fulfill potentials in the world, while the spiritual endeavor is to liberate from suffering, transform the mind, and dissolve ego boundaries (Turnbull, 2014). It is an individual choice regarding which endeavor they are committed to, since spiritual principles and disciplines need to evolve out of personal lived experiences (Progoff, 1980). Only when they fit the current context of an individual life can they contribute to well-being.

Limitations and Future Research

Some limitations of the current meta-analysis should be acknowledged. First, to confirm a mechanism of action, changes in decentering should precede changes in psychological problems. This temporal antecedence is necessary to establish causal relationships. However, the current study used correlational data and cannot demonstrate the temporal sequence of changes in decentering and psychological outcomes, which precluded strong inferences about the causality of the relationship observed. Admittedly, correlational analysis is restricted by the lack of random assignment and of temporal precedence which are necessary conditions to establish causal chains (e.g., Fiedler et al., 2011; Rucker et al., 2011). Thus, we refrain from using terms such as mediation or mechanism in drawing conclusions. Instead, the result could be interpreted as that the correlation between mindfulness and decentering carries over to alleviate psychological problems. Understanding this correlation is important to uncover the psychological processes leading from mindfulness to psychological outcomes.

The second limitation was that the current study did not examine how each facet of mindfulness differed in the direction and strength of its association with decentering and with psychological problems. There was evidence that non-reactivity to inner experiences was negatively associated with the activation of insula involved in automatic emotional responses (e.g., Paul et al., 2013) and inversely associated with depression and anxiety symptoms (e.g., Desrosiers et al., 2013). In contrast, the observing facet of mindfulness was positively associated with psychological symptoms in non-meditators (e.g., Baer et al., 2006) and anxious arousal in treatment-seeking adults (e.g., Desrosiers et al., 2013), and negatively associated with other facets of mindfulness (e.g., Brown et al., 2015). One possible explanation was that observing needed to work in tandem with a non-judging attitude, which can be properly trained through mindfulness practice (e.g., Biehler & Naragon-Gainey, 2022; Brown et al., 2015). Given these mixed findings, more research is needed to investigate the effects of each facet of mindfulness and of decentering on particular conditions (e.g., physical, psychological) and symptoms (e.g., depression relapse, generalized anxiety). A fine-grained analysis of facets could inform interventions to match specific components with individual needs and target at training essential components for specific populations such as the elderly, patients with chronic illness, and minority groups (Fresco & Mennin, 2019). With this in mind, future studies need to develop psychometrically validated scales indexing specific components of mindfulness and decentering (Grossman & Van Dam, 2011). Alternative methods such as ecological momentary assessments (e.g., dairy, mobile apps) can capture changes in real-time, real-world environments, and provide a clear picture of how people calibrate their coping strategies to changing circumstances in everyday life (e.g., Biehler & Naragon-Gainey, 2022; Shoham et al., 2018).

Relatedly, a promising direction for research is to further differentiate the construct of decentering and explore whether decentering has interpersonal benefits. The current conceptualization of decentering comprises three facets: meta-awareness, disidentification, and non-reactivity (e.g., Bernstein et al., 2015). However, the appropriate practice of decentering is grounded in the thorough understanding of the true nature of all phenomena, including ourselves (e.g., Grabovac et al., 2011; Van Gordon & Shonin, 2020). According to traditional Buddhist teachings, decentering involves observing and seeing oneself, forgetting oneself, and recognizing oneself together with the surrounding context (Gyatso, 2000). As such, we are able to attune to the common ground with fellow men (e.g., yearning for well-being) and shift attention away from excessive self-concerns (e.g., Soler et al., 2021). Thus, another core attribute of decentering should be noted, that is, decentralization of the ego.

It is not the mind’s desire for particular objects so much as its obsession with self-validation that perpetuates the seeking (Gethin, 2015). Where Buddhist psychology and Western psychology diverge is that Western psychology solidifies the central position of self who creates a cohesive self-concept and Buddhism views no-self as the ultimate organizing principle of life that liberates us from self-centric illusion (Rubin, 1996). What does this difference inform us? It speaks to a problem when our live gravitates around this central position of the conceptualized “I.” The consequence is an objectified me in the eyes of an egocentric I who judges oneself by the extent to which one has achieved success or accumulated wealth, but at the expense of a persisting sense of lack and defect in our life (Brown, 2015; Ryan & Rigby, 2015). In fact, the Buddhist doctrine of no-self does not negate the agency of self in performing acts, thinking ideas, and feeling emotions. Instead, decentering from egocentric perceptions of experiences culminates in a drastic change in the sense of self following deep states of mediation, such as a positive self-regard, self-confidence, and acceptance of oneself (Hölzel et al., 2011; Soler et al., 2021). At a more advanced level of meditation practice, the self is conceived as part of a larger whole and becomes whole when in communion with other beings (Dambrun & Ricard, 2011).

The point is not to renounce the subjective sense of self or to glorify the sense of no-self, but to take a dialectical approach in human existence. The co-existence of both views of selfhood is necessary for optimal functioning because they serve specific purposes in particular contexts (Rubin, 1996). For example, when practicing the craft of dancing, we become our experiences without excessive self-consciousness. A consolidated, agentic self is the early developmental stage of disengaging from self-preoccupation, and they are not irreconcilable ways of being (Engler, 1984). In other words, one’s self has to be built before one can move beyond it. Recent meta-analyses showed that decentering induced an experience of transcendence which was characterized by a profound sense of union with others and with nature (e.g., Hanley et al., 2023; Sanyer et al., 2023). Experimental studies provided evidence that when university students were given decentering instructions, they were less inclined to interpret the intention of others as hostile or aggressive in ambiguous social situations, compared with their counterparts who were instructed to immerse in their thoughts (e.g., Schans et al., 2020). Among a community adult sample who received a 6-week mindfulness training course, the improvement in mindfulness predicted greater senses of social connection, which was mediated by an increase in decentering (e.g., Adair et al., 2018). However, it remains unclear if disidentification from mental content is a sufficient condition to produce interpersonal benefits.

Arguably, decentering involves not only keeping a distance from inner states, but also loosening the grip of egocentric concern. Decentralization of the ego offers an alternative way to engage with the world. Mindfulness meditation provides an opportunity to develop the flexibility of shifting between alternative ways of being and relating to our life. Those moments of surrendering the self-centered positioning give us a glimpse of what lies on the other side of the self-other division: being at one with our core self and other beings in a deeper, tender way. Thus, a refined conceptualization of decentering is needed to test the hypothesis that decentering at the base of mindfulness can extend the sense of self and facilitate a progression towards a higher sense of interrelatedness. Clarifying the different dimensions of decentering and the hindrances to effective decentering can shine a light on the refinement of measurement instruments as well as of prevention and intervention programs.