Studying self-related processes is necessarily complex on account of involving the notion of self, which has long been recognized as a “mushy, muddle-headed construct without empirical referents” that encompasses multiple meanings (Westen, 1992) (p. 4). Indeed, various disciplines employ multiple terms in reference to similar conceptions of self and also employ the same (or similar) terms to refer to different SRPs –a conceptual morass known as the jingle-jangle fallacy (Roeser et al., 2006). Here we attempt to define and delineate various SRPs before querying how they are impacted by MBIs. SRPs have been defined as “processing requiring one to evaluate or judge some feature in relation to one’s perceptual image or mental concept of oneself,” which the authors differentiate from embodied, pre-reflective self-specifying processes “that specifies the self as subject and agent by implementing a functional self/non-self distinction”(Christoff et al., 2011b, p.104). However, the relationship between some pre-reflective self-specifying processes and evaluative self-related processes may less distinct and more dynamic (Balconi, 2010; Gallagher and Daly, 2018). For the sake of simplicity, in this review we use SRPs in a more general sense to identity a range of potential mechanisms of MBIs, whether conceptual and evaluative processes, pre-reflective embodied processes, or processes of self-regulation. While recognizing that there remain many debates about to delineate and differentiate them (Christoff et al., 2011a; Northoff, 2011), we organize the SRPs investigated in this evidence map along a continuum from more embodied “self-as-subject” processes to more conceptual “self-as-object” processes (Figure 1).

Perhaps the most fundamental aspect of the self-as-subject is the perspectival self, or the sense that our experience is situated from a specific, first-person perspective or vantage point (Zahavi, 2005, 2012). Other terms that have been used for the more basic embodied senses of self include core self, proto-self (Damasio, 2010) and minimal self (Gallagher, 2000). The broader concept of minimal self often includes additional pre-reflective processes central to embodiment. Interoception, or body awareness, is supported by the sense of ownership, or the extent to which thoughts, emotions, and sensations present themselves as “me” or “mine” vs. “not me” or “not mine.” The sense of agency – or self as agent, doer, or controller – can be described as the sense of ownership extended to actions (Gallagher, 2000), although agency and ownership can also be empirically differentiated (Tsakiris et al., 2007).

In contrast to the pre-reflective self-as-subject, the upper circle in Figure 1 represents senses of self supported by conceptual, narrative, and evaluative processes. The narrative self is often described as the autobiographical story of me as a person with a temporally extended identity, including the conception of past and future selves, multiple imagined selves (Gallagher, 2000). The narrative self also includes various social identities, group memberships and associated roles. Many key psychological constructs such as self-esteem and self-evaluation could be considered more self-as-object SRPs. Negative self-evaluations include self-criticism and rumination, whereas positive evaluations include self-esteem, self-worth, and self-praise. Many forms of psychotherapy focus on decreasing rumination, increasing self-esteem, and shifting the valence of the self-concept from negative to positive (Kyrios et al., 2016).

Lastly, we identify SRPs that function as self-regulation skills, specifically decentering and self-efficacy. Self-efficacy, for example, is the perceived capability of an imagined future self to perform a target behavior (Bandura, 1977) and is a type of self-evaluation that relates not only to imagined qualities, capacities, and goals of the conceptual self or identity, but also to the embodied and agentic self and its capacity to engage in action – especially goal-related behavior. As such, we represented these SRPs at the intersection of bodily senses of self and conceptual senses of self (where the two circles overlap in Figure 1).

Mindfulness-based interventions operate at the confluence of multiple disciplines, including clinical, health, cognitive, and positive psychology, as well as Buddhist theories of mind. As products of clinical and health psychology, MBIs might be expected to promote any number of SRPs associated with self-concepts and self-regulation. However, in line with the Buddhist theories from which they are derived, many contemporary models for the mechanisms of mindfulness also propose that MBIs affect (e.g., attenuate or transform) many, if not most, dimensions of selfhood, not just conceptual ones (Hölzel et al., 2011; Teasdale and Chaskalson, 2011a, b; Vago and Silbersweig, 2012; Kabat-Zinn, 2013; Brown et al., 2015; Lutz et al., 2015). Indeed, many theories hypothesize that changes in more embodied self-specifying processes, including ownership, agency, and interoception underlie the beneficial effects of mindfulness meditation (Olendzki, 2006; Kabat-Zinn, 2013; Brown et al., 2015; Hadash et al., 2016). However, exactly which aspects of the self are being transformed, attenuated, or transcended through meditation have been subject to considerable debate in historical sources across Buddhist traditions (Albahari, 2002), as well as in contemporary humanistic and experimental contexts (Hölzel et al., 2011; Vago and Silbersweig, 2012; Ostafin et al., 2015; Yaden et al., 2017; Desbordes, 2019). The following evidence map introduces a comprehensive framework the different types of SRPs that have been investigated in MBIs, describes how they are theorized to function as mechanisms, and evaluates the strength of empirical evidence for SRPs target engagement and target validation.

Self-related processes are theorized as key mechanisms of MBIs. According to the experimental medicine approach, hypothesized intervention targets must meet several other basic criteria to qualify as a mechanism (Kazdin, 2007; Onken et al., 2014; Riddle and Science of Behavior Change Working Group, 2015). In addition to theory (target identification), validated measures or assays that accurately represent the theorized SRP construct must exist or be created. Once created and validated, the SRP assay must be reliably engaged by mindfulness meditation training (target engagement), and the degree of SRP target engagement must account for the change in desired outcomes (target validation).

The current paper is an evidence map that uses a multidimensional methodology to synthesize the evidence. An evidence map is generally a type of knowledge synthesis that maps key concepts, types of evidence, and gaps in knowledge. While unrestricted to a specific methodology, depending on goals and resources an evidence map typically includes: a review of key documents; consultation with stakeholders or experts; specific inclusion criteria and a systematic method of search; data extraction; and critical evaluation (Miake-Lye et al., 2016; Snilstveit et al., 2016). Compared to other types of reviews, evidence maps, or more pointedly “evidence gap maps” are particularly well-suited to “identify broad trends in a particular area of research and highlight key shortcomings in the existing evidence base” (Snilstveit et al., 2016, p. 123). One defining feature of an evidence (gap) map is the visual representation evidence in an accessible, user-friendly figure or graph in order to most effectively highlight “absolute gaps where there are few or no studies…where new primary studies or systematic reviews can add most value” (Miake-Lye et al., 2016; Snilstveit et al., 2016, p. 123).

A detailed description of our methodology can be found in the Supplementary Materials (see Supplementary Methods) but is described briefly here. We initiated and anchored the evidence map with a systematic review of target engagement in RCTs of standard MBIs since RCTs present the strongest evidence for causal inference of the effects of mindfulness meditation training on specific outcomes. The target identification phase was performed by an expert panel of five mindfulness researchers with broad knowledge of SRPs. In addition to search terms beginning with or including the word “self,” the panel generated 30 SRP constructs to serve as search terms, namely: agency, body awareness, decentering, disidentification, ego, embodiment, experiential self, identity/identification, interoception, meta-awareness, narrative self, observing self, perceived control, reperceiving, self-awareness, self-compassion, self-criticism, self-distancing, self-efficacy, self-esteem, self-loathing, self-monitoring, self-praise, self-referential processing, self-regulation, self-related rumination, self-worth, sense of agency, sense of control, and sense of ownership.

Eligible studies were English-language RCTs of standard Mindfulness-Based Stress Reduction (MBSR) and Mindfulness-Based Cognitive Therapy (MBCT) with one or more active and/or inactive control condition(s), with adult participants (age 18 or older) from both clinical and non-clinical samples. Of the 10,802 studies identified, only 14 standard MBI RCTs assessed SRPs. In addition, only five of the 30 SRP constructs generated by the expert panel had been measured within an RCT of a standard MBI, namely: self-evaluation, self-compassion, rumination, self-esteem, and self-efficacy.

Because the initial review demonstrated that very few SRPs had been measured in the context of standard MBI RCTs, we adapted our mapping approach to match the inchoate state of the research. The expert panel agreed to specifically include (and allow less rigorous inclusion criteria for) SRP constructs that have been theorized as central or key mechanism of MBIs, specifically interoception, decentering, selflessness and self-transcendence (Hölzel et al., 2011; Vago and Silbersweig, 2012). This allowed us to provide both an evidence map and an evidence gap map for a wider range of SRPs than yielded by the initial review.

Thus, using the experimental medicine framework (Kazdin, 2007; Onken et al., 2014; Riddle and Science of Behavior Change Working Group, 2015) we created an “evidence gap map” (Snilstveit et al., 2016) by describing other mindfulness meditation studies that investigate SRPs and organized them hierarchically according to strength of evidence. The highest level of evidence are meta-analyses of multiple MBI RCTs where an SRP fully mediates desired outcomes (Kazdin, 2007), although few SRPs met this criteria (Gu et al., 2015; van der Velden et al., 2015; Alsubaie et al., 2017). If no meta-analyses of MBI RCTs were available, then single RCTs of standard, then modified MBIs were described. If the target had not been tested in a longitudinal RCT, then brief experimental meditation induction studies were permitted. If induction studies were absent, then cross-sectional studies of meditators vs. non-meditators were reviewed.

The review of each construct included four parts: (1) a definition and how it is theorized to function as a mechanism of MBIs (target identification); (2) description of target engagement by mindfulness meditation training according to a hierarchical levels of evidence (meta-analysis of MBI RCTs > MBI RCTs > modified MBI RCTs > induction studies > cross-sectional studies); (3) description of target validation, or the degree to which target engagement account for desired outcomes, using similar levels of evidence above, with Kazdin’s full mediation criteria at the top (Kazdin, 2007); and (4) commentary to discuss caveats and other limitations. Figure 2 provides a user-friendly visual presentation of the evidence and evidence gaps, which are further detailed in the following sections. A table of all SRP studies and constructs can be found in the Supplementary Material (see Supplementary Table 1).

Conceptual SRPs constituted the highest level of evidence as mechanisms of MBIs. Five SRP constructs were organized under two types of self-evaluation: negative and positive. Specifically, rumination, dysfunctional attitudes and negative automatic thoughts were categorized under negative self-evaluation, while self-compassion and self-esteem were categorized under positive self-evaluation.

In this evidence map review, the researchers determined that the measurement of SRPs is well defined for some constructs, whereas other constructs have limited measurement tools and extant assays for some SRPs are surrounded by controversy. As such, we provide the reader with guidance about possible ways to measure SRPs. Table 1 provides a summary of commonly used SRP measurement tools in MBIs. To our knowledge, well-validated assays exist to measure the SRPs of rumination, dysfunctional attitudes, and negative automatic thoughts. However, it should be noted that most rumination scales contain a negatively valenced self-evaluation dimension (brooding) and a neutral or non-valenced self-focus dimension (reflection). It is the negatively valenced dimension (brooding) and not reflection that is both reduced by MBIs and mediates reductions in depression (Shahar et al., 2010; Armstrong and Rimes, 2016). In contrast, MBI studies that combined the two scales often found no effect (Robins et al., 2012; Eisendrath et al., 2016). In other words, it is the negative self-evaluation dimension of rumination rather than self-focus alone that explains the beneficial effects of MBIs on depressive symptoms.

Additionally, validated assays exist to measure decentering, self-compassion, self-efficacy, and self-esteem. The Self-Compassion Scale (SCS; Neff, 2003) is the most widely used scale for measuring self-compassion, but it has received some criticism over its construct validity and its status as a therapeutic mechanism (Muris and Otgaar, 2020). Half of the scale is comprised of “uncompassionate self-responding” reversed items that are highly correlated with internalizing psychopathology (anxiety and depression). Thus, use of the SCS total score could obscure what processes are occurring in therapy, and inflate the relationship with psychopathology. In rebuttal, Neff acknowledged that the two sides of the scale (i.e., compassionate and uncompassionate self-responding) are distinct but related, and presented empirical data to defend the validity of the SCS (Neff, 2016, 2020). Nevertheless, a number of new self-compassion-related scales have been developed (Kraus and Sears, 2009; Gilbert et al., 2011; Gu et al., 2020), but have yet to be widely utilized in MBI research (Kraus and Sears, 2009; Gilbert et al., 2011; Gu et al., 2020).

Many mindfulness scales, including the Five Facet Mindfulness Questionnaire (FFMQ) (Baer et al., 2006) and the Freiburg Mindfulness Inventory (FMI) (Walach et al., 2006), contain items related to decentering. However, Bernstein et al. (2015) identified two assays that specifically target decentering within MBIs: the decentering subscale of the Experiences Questionnaire (EQ; Fresco et al., 2007) and the Toronto Mindfulness Scale (TMS; Lau et al., 2006), which was based on an operational definition of mindfulness by Bishop et al. (2004) related to non-elaborative attention characterized by curiosity, acceptance, and openness to experience, with all items referring to an immediately preceding meditation session. Decentering is plagued by imprecision and issues with construct validity (Bernstein et al., 2019). At least fifteen different terms are in use to refer to decentering-like processes including cognitive defusion, cognitive distancing, decentering, detached mindfulness, disidentification, metacognitive awareness, metacognitive mode, observing self reperceiving, self-as-context, self-distancing, third-person perspective, and witness consciousness (Deikman, 1982; Masuda et al., 2004; Bernstein et al., 2015). While many of these terms are used interchangeably, it is unclear to what extent the terms and their associated assays measure the same or different constructs. For example, while both the TMS and EQ measures for depression purportedly measure “decentering,” they are not correlated with each other, and appear to measure different constructs (jingle fallacy) (Bieling et al., 2012). The jangle fallacy also appears to be at play, as constructs with different names, such as mindfulness and decentering, are more highly correlated (r = 0.81) than constructs with the same name (Carmody et al., 2009). In addition, some items on the decentering scales, e.g., “I am better able to accept myself as I am” may better reflect self-compassion and self-acceptance,” than disidentification (Fresco et al., 2007).

The Multidimensional Assessment of Interoceptive Awareness (MAIA; Mehling et al., 2012, 2018) is one of the main self-report measures of interoceptive attention that is used in mindfulness research. Yet, other facets of interoception are less readily measured. Interoceptive accuracy has often been assessed via assessments of individuals’ experiences of various body areas or bodily functions (e.g., heartbeat counting). However, these tasks’ validity has been criticized by psychophysics experts (Brener and Ring, 2016; Ring and Brener, 2018). A number of other objective performance measures have been developed to assess interoceptive accuracy in different bodily systems, including measures of respiratory tracking, blood glucose, tactile sensitivity and detection, joint position and kinesthesia, proprioceptive drift, and arousal as measured through heartbeat and skin conductance (see Treves et al., 2019).

Three main validated measures have been developed for the assessment of self-transcendence in the context of meditation. The Non-dual Awareness Dimensional Assessment-Trait (NADA-T; Hanley et al., 2018) is a well validated self-report measure that was created through a factor analysis of preexisting mysticism and meditation scales. It contains two subscales: self-transcendence and bliss. The self-transcendence subscale includes multiple items pertaining to self-world boundaries such as “I have had an experience in which the boundaries of my self dissolved,” but also includes items that may be tracking other kinds of self-related changes, such as “I have experienced all notion of self and identity dissolve away” and “I have experienced the insight that ‘all is one.”’ Also embedded within it are measures not directly related to SRPs; the bliss subscale measures positive affect experiences, such as the item “I have experienced an all-embracing love.”

The Perceived Body Boundaries Scale (PBBS; Dambrun, 2016) is a single-item measure developed to assess “the sense of being more or less separated from the world” (p. 91). The PBBS uses a series of images of human forms with increasingly transparent boundaries to indicate gradations of body boundary salience, with the intention of measuring the boundaries of the self, which may or may not align with the boundaries of the physical body (Dambrun, 2016). Lastly, the Spatial Frame of Reference Continuum (SFoRC; Hanley and Garland, 2019) is a single-item measure through which participants rate how far their sense of self expands into the world. The scale uses a single image of a human form surrounded by concentric circles that expand outward. Similar to decentering and mindfulness, measures of self-transcendence and selflessness have issues with construct validity and jingle-jangle fallacies. Different terms and scales are being used to measure largely overlapping constructs. In addition, the identical terms “selflessness” and “self-transcendence” are used both to refer to a trait-based social identity construct about the degree to which other people are included in one’s self-concept (Figure 1, upper circle), as well as to refer to state-based embodied SRPs concerning self-world boundaries (Figure 1, lower circle).

This knowledge synthesis investigated the potential role of self-related processes (SRPs) as a mechanism of change in standard mindfulness-based interventions (MBIs). A change or attenuation in selfhood is frequently advanced as a mechanism of MBIs, a perspective that comes at least in part from the substantial influence of the theory and practice of Buddhism on MBIs. However, which SRPs should be considered the target is often un- or under-specified. Contemporary research across multiple fields has demonstrated that the sense of self is grounded in multiple processes, ranging from pre-reflective embodied “self-as-subject” processes to conceptual and often valenced evaluative “self-as-object” processes. This evidence map has attempted (1) to synthesize and clarify which and in what way these processes are theorized to be mechanisms of MBIs, (2) to evaluate the current degree of evidence across the full range of SRPs, and (3) to indicate where research into SRPs require additional clarification and precision in order to navigate the jingle-jangle fallacy.

Based on the current review, extant data show that MBIs significantly engage some SRPs, regardless of MBI or population type, which confirms that some SRPs have a signal as potential mechanisms leading to beneficial health outcomes. Most of the SRPs that have been measured in MBI RCTs are conceptual senses of self or self-regulation skills. Specifically, the SRPs measured and engaged by MBIs include increased positive and decreased negative self-evaluation, as measured by rumination, negative thoughts and attitudes, self-compassion, self-esteem and self-efficacy. With support from multiple meta-analyses, rumination showed the strongest evidence as an MBI mechanism for depression, with other physical and mental health outcomes also supported. Self-compassion showed consistent target engagement but was inconsistently related to improved health outcomes. Self-efficacy and self-esteem were reliably engaged by MBIs but have not been evaluated as mechanisms. Evidence for decentering as a mechanism was mixed and undermined by inadequate construct validity and discrepancies between theory and assay. Across all RCTs, MBIs engaged SRPs more than passive, but not active controls, which suggests that engagement of SRPs is neither specific to nor greater in MBIs. Overall, positively and negatively valenced conceptual senses of self investigated through clinical and health psychology have received the most attention and have the larger evidence base for a mechanistic relationship with MBIs.

In contrast, very few, embodied SRPs have been measured in an MBI or longitudinal intervention study. Interoceptive processes were the only process with enough studies of any kind to facilitate a meta-analysis. Evidence for interoceptive processes as mechanisms of MBIs was mixed, and dependent on type of interoceptive process and method of measurement. Aside from interoception, measurement of other embodied self-specifying processes were entirely absent from standard MBI trials, and with the exception of one non-standard MBI trial, came mostly from experimental induction studies in healthy samples or qualitative case reports. In addition to the lack of data from clinical trials, evidence for attenuations in self-specifying process as mechanisms of MBIs or meditation is limited by measurement issues and lack of assessment of enduring outcomes.

While this evidence (gap) map is one of the first attempts to identify and evaluate a wide range of SRPs in MBIs, it also has many limitations. The current review may not be exhaustive, as SRPs are often measured as secondary outcomes and thus some important studies may not have been identified in this map. In addition, this review investigated the effects of meditation practice on SRPs without differentiating the types of meditation practices, which may impact target engagement and target validation (Cullen et al., 2021). Additionally, our review depended on the measurement tools used in extant research, which have some limitations. The use of well-validated SRP measures is imperative to advancing this field.

As one of the first reviews to integrate and evaluate evidence for SRPs as mechanisms in MBIs, many other questions still remain to be explored. Importantly, this evidence map suggests that greater attention needs to be placed on identifying, delineating and measuring specific SRPs. Theoretical models proposing changes in sense of self have often insufficiently specified SRPs, leaving ambiguity as to whether those processes are conceptual and evaluative “self-as-object” SRPs, pre-reflective and embodied “self-as-subject” SRPs, or self-regulation skills. They have also insufficiently attended to whether increases or attenuations of SRPs is the proposed mechanism of change. This evidence (gap) map found evidence that mindfulness training leads to decreases in certain conceptual senses of self (negative self-evaluation) and increases in others (positive self-evaluation) suggesting an important and often underemphasized role of valence. Although preliminary, current research also suggests that mindfulness training may lead certain embodied processes (e.g., sense of body boundaries) to decrease while others (e.g., interoception) increase, although how these changes are related to outcomes is still unknown. Attending to these differences remains crucial for both theoretical models and empirical research, as certain directions of change in specific SRPs have better evidence for target engagement and target validation than others, and researchers should not expect that a global attenuation of SRPs will be optimal in all circumstances (Britton, 2019).

Given that different meditation practices are intended to target different forms of selfhood in different ways, more attention should be paid to practice-specific effects (Dahl et al., 2015; Lutz et al., 2015). Furthermore, given that the current evidence emphasized novices in MBIs, with data from advanced Buddhist meditators being included only when data from MBIs was unavailable, it is possible that certain additional SRPs might change as a result of years of mindfulness practice rather than in the course of a typical 8-week MBI. However, given that the review focused on mechanisms of MBIs, it is unclear whether SRPs engaged through a longer commitment to mindfulness practice should be considered mechanisms of change for MBIs. Finally the social self – or the self in relation to others, or as part of a group – was only briefly discussed in the section on self-transcendence. Given that many forms of self-hood, both conceptual and embodied, can be influenced by both culture and group membership (Canby et al., 2020), the impacts of MBIs through social dimensions of self deserve more attention.

The findings of the current review has several implications for mindfulness-based interventions. The strongest evidence for an SRP-based mechanism of MBIs’ beneficial effects comes from increasing positively valenced and/or decreasing negatively valenced conceptual senses of self. Consistent with other studies, this suggests that mindfulness-related wellbeing is better explained and achieved by self-concept enhancement than Buddhist-derived ego-quieting or multi-level SRP reduction (Gebauer et al., 2018). In order to maximize efficacy, those delivering MBIs may want to consider how self-concept enhancement can be emphasized in the curriculum.

Mindfulness-based intervention research on embodied SRPs remain understudied, especially in relation to outcomes. However, there is reason to think that reductions in embodied SRPs or global reductions in all SRPs is not the primary mechanism of MBI benefits. In a recent MBI trial, “self-disturbances” were among meditation-related adverse effects associated with enduring impairment (Britton et al., 2021). In addition, a qualitative study of meditation-related changes in sense of self found that reports of more global reductions in SRPs was associated with higher levels of functional impairment (Lindahl and Britton, 2019). Furthermore, attenuations in embodied SRPs are also associated with severe psychopathology, including psychotic disorders and dissociation (Sierra and David, 2011; Sass, 2014; Borda and Sass, 2015; Medford et al., 2016). Thus, strategies for decentering or “disidentification” from certain aspects of experience should perhaps be circumscribed to negative self-evaluation.

In addition to mental health risks, medical ethicists have identified epistemic costs and raised ethical concerns that MBIs are “metaphysically loaded with philosophical claims about the self” that may create tension with people’s core beliefs (Ratnayake and Merry, 2018 p, 567). The authors also point out that the metaphysical commitments of mindfulness practices means the claim that they are “universal” is “problematically misleading” (p. 567). MBIs can assuage these concerns, maximize benefits and minimize harms by basing intervention components and practice choices on the available evidence base rather than on “metaphysically loaded” exercises and theories that are not substantiated by the evidence base.