Introduction

There has been a recent increase in research focused on empathy, compassion, and prosocial behaviors (Kirby 2017; Strauss et al. 2016). Although there are varying definitions of empathy and compassion, they are often considered related but distinct prosocial emotions that consist of cognitive and affective components and can be learned with practice (Bibeau et al. 2016; Goetz et al. 2010). Empathy involves vicariously experiencing another’s emotions by recognizing, understanding, and resonating with their emotional state (“putting yourself in someone else’s shoes”; Hogan 1969; Lazarus 1991; Strauss et al. 2016). Compassion takes empathy a step further and involves not only emotional recognition, understanding, and resonation but also the ability to tolerate one’s own emotional reaction and the motivation to act to relieve the others’ suffering (“suffering with”; Gilbert 2010; Strauss et al. 2016). Actions taken with altruistic intentions to help or benefit another person are broadly considered prosocial behaviors (e.g., volunteerism, charitable donation, care-taking; Penner et al. 2005). Research supports the idea that greater empathy leads to greater compassion, and greater compassion leads to greater prosocial behavior (Lim and DeSteno 2016).

Prosocial emotions and behaviors are important for both individual and societal well-being. Empathy and compassion are emphasized across diverse social institutions, including healthcare, education, and justice systems, as well as most world religions (Faulkner and McCurdy 2000; Goetz et al. 2010). They are thought to confer adaptive evolutionary value by guiding individuals to protect and care for their offspring, family, as well as other community members, thereby maximizing the likelihood of survival and genetic propagation (Goetz et al. 2010). Prosocial outcomes have a positive public health impact because they not only benefit the individual receiving help, but they also benefit the helper. Indeed, a large body of research demonstrates that engaging in prosocial behavior is associated with greater happiness and psychological well-being, indices of physiological health (e.g., increased heart rate variability, immune function, telomere length, genetic expression), better physical functioning, better interpersonal relationships, and decreased morbidity in medical populations (Dunn et al. 2008; Hoge et al. 2013; Ironson 2007; Nelson et al. 2016; Pace et al. 2009; Weinstein and Ryan 2010). These benefits are greater for prosocial behavior as compared to self-focused helping behavior (e.g., Nelson et al. 2016). Given the wide range of social problems currently harming individuals and societies worldwide, the need for greater empathy, compassion, and prosocial behavior is clear (Hurst et al. 2016; Pascoe and Richman 2009).

Meditation is one way to increase an individual’s empathy, compassion, and prosocial behavior. Meditation encompasses a collection of mental training practices that involve self-regulating one’s attention toward a chosen object of awareness from one moment to the next; it can take various different forms depending on how and where attention is focused (Kabat-Zinn 1982; Walsh and Shapiro 2006). Meditation practices have been used for centuries across a range of contemplative communities and historically emphasized as methods to reduce suffering for the self and others within a moral or religious context of benevolence and nonharming (Goldstein and Kornfield 2001; Nydahl 2008; Sears et al. 2011). Over the past twenty years, meditation practices have been increasingly secularized and integrated into psychological interventions to improve both negative and positive emotional outcomes (Kirby 2017).

Two meditation practices that have received particular attention are mindfulness meditation and loving kindness meditation (LKM) practices derived from Buddhist contemplative traditions. Mindfulness meditation involves self-regulating one’s attention to intentionally notice present moment experiences openly and nonjudgmentally as they occur (Sears et al. 2011). It incorporates the related practice of concentration meditation in that it involves focused concentration on an object of experience in the present moment. LKM is a more directly prosocial meditative practice aimed at increasing four specific other-oriented positive attitudes: loving kindness, compassion, empathic joy, and equanimity. LKM practices involve intentionally cultivating awareness of feelings of warmth, kindness, and compassion for others through mental visualizations, mantras, and/or other aspirational phrases (Wallace 1999). There are also compassion meditation practices, which can be similar to LKM practices but have a unique focus on imagining another’s suffering and relieving that person’s suffering (e.g., by extending a heartfelt wish or imagining a golden beam of light toward them). Movement-based meditation practices derived from disciplines such as yoga and tai chi, which combine mindfulness meditation with physical postures or exercises, have also received increased research attention (Luberto et al. 2013).

There is a strong evidence base to support the efficacy of meditation-based interventions for improving emotional outcomes. This research work had initially been focused on decreasing negative emotions (i.e., rather than increasing positive emotions) using mindfulness-based interventions such as mindfulness-based stress reduction (MBSR; Kabat-Zinn 1982) and mindfulness-based cognitive therapy (MBCT; Segal et al. 2012). The results of several systematic reviews and meta-analyses of mindfulness-based interventions suggest that these treatments significantly improve stress, anxiety, depression, quality of life, and emotion regulation across a range of psychiatric and medical populations (Bohlmeijer et al. 2010; Eberth and Sedlmeier 2012; Gotink et al. 2015; Hofmann et al. 2010; Khoury et al. 2013; Piet et al. 2012). Reviews of movement-based mindfulness practices also show promising results for improving emotional problems (e.g., anxiety, depression), though these results are more preliminary given the limited methodological quality of these studies to date (Kirkwood et al. 2005; Luberto et al. 2013; Uebelacker et al. 2010).

More recently, research has begun to focus on LKM practices to decrease negative and promote positive emotions. Hofmann et al. (2010) suggested that LKM practices may be integrated into cognitive-behavioral therapies to improve emotions and behaviors related to interpersonal relationships, and a recent meta-analysis found that LKM indeed improves depression, mindfulness, compassion, self-compassion, and positive affect (Galante et al. 2014). Other meta-analyses of LKM for improving self-oriented positive emotions (Zeng et al. 2015) and general psychosocial outcomes (Shonin et al. 2015) have shown significant benefits. A narrative review also suggested that compassion meditation promotes prosocial outcomes in psychotherapists (Bibeau et al. 2016 ).

Despite the multiple reports of meditation and emotional well-being, no research has systematically reviewed the results of meditation interventions for prosocial outcomes. Previous systematic reviews and meta-analyses have tended to focus on one specific type of meditation practice (e.g., mindfulness or LKM; Galante et al. 2014; Zeng et al. 2015), negative emotions (Hofmann et al. 2010), or self-focused positive emotions (e.g., Zeng et al. 2015). Those that did incorporate empathy and compassion outcomes either did not specifically include prosocial search terms (Galante et al. 2014; Shonin et al. 2015) or were not systematic and only examined outcomes in one specific population (i.e., psychotherapists; Bibeau et al. 2016). Thus, a comprehensive and systematic review of meditation for prosocial outcomes is lacking.

The purpose of the current study is therefore to conduct a systematic review and meta-analysis of randomized controlled trials of meditation-based clinical interventions for improving prosocial emotions and behaviors. Specifically, the aims are to synthesize existing results regarding effects and potential mechanisms of meditation for prosocial outcomes, estimate the effect size of meditation on prosocial outcomes, assess the quality of trials conducted, identify directions for future research, and draw evidence-based conclusions to guide future research and clinical practice.

Method

Literature Search

A literature search was performed by a medical librarian (LP) in the Ovid Medline, PubMed, Ovid PsycINFO, CINAHL, Embase, Cochrane Library, and ClinicalTrials.gov databases from inception through April 2016. Similar to previous reviews of meditation (Gotink et al. 2015), search terms were intended to capture studies of meditation interventions that have been secularized for delivery in standard clinical practice settings. We only included secular practices because these are more likely to be offered in standard clinical practice settings (e.g., MBCT, MBSR), they can promote a wider outreach for individuals who may not subscribe or feel comfortable with nonsecular practices, and much of the literature to date has tended to focus on secularized interventions. Nonsecular practices are also often religion-specific and may not be generalizable. Prayer was excluded as it is inherently nonsecular. Also similar to previous reviews, cognitive-behavioral therapies (CBT) that do not use formal meditation practice consistently as the foundation of treatment were excluded (e.g., traditional CBT, dialectical behavior therapy, acceptance and commitment therapy; Hofmann et al. 2010). Thus, search terms included the following: meditation, mindfulness, MBSR, MBCT, mind-body therapies, tai chi, yoga, empathy, compassion, sympathy, love, altruism, and kindness. Each search query was combined with a filter based on Royle and Waugh’s search strategy for identifying randomized controlled trials for systematic reviews (Royle and Waugh 2005). An additional filter was used to limit to English language studies. No publication date limits were used. See Appendix for the full search strategy in Ovid Medline.

Eligibility Criteria

Randomized controlled trials of a meditation-based intervention that assessed at least one quantitative outcome related to prosocial emotions or behaviors were eligible for inclusion. Meditation-based interventions were considered those whose theoretical foundation incorporated philosophies from meditative traditions and provided direct and consistent training in meditation practices as the primary foundation of the intervention (i.e., across at least half of the sessions). Studies that only assessed self-focused compassion were excluded. Unpublished manuscripts, conference presentations, and dissertations were excluded. Non-English studies were excluded due to insufficient funds for translation. We did not exclude studies based on patient demographics such as age or clinical status (i.e., studies of children and adults of any population were included).

Data Extraction and Synthesis

Two independent reviewers (CML and NS) extracted data from each study and discussed results to ensure agreement. Any discrepancies were resolved through discussion with the senior author (GY). The following data were extracted: study sample, intervention type and format, control group type and format, intervention dose and adherence, prosocial outcome variables and time points, and results for effects on prosocial outcomes. We also extracted any reported data on potential mechanisms of meditation effects (e.g., mediation or correlation analyses examining relationships between changes in prosocial outcomes and other biopsychosocial variables).

Risk of Bias Assessment

Two independent reviewers (CML and NS) assessed risk of bias for each included study according to Cochrane Collaboration guidelines (Higgins and Green 2008). Risk of bias was assessed as high, low, or unclear for each of the eight domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessors, incomplete data, selective reporting, baseline imbalance, and differential attrition (Jüni et al. 2001; Liberati et al. 2009). In our synthesis, particular attention was paid to low-risk studies, defined as studies with low risk on most (5 out of 8) of the domains assessed. Studies that were not deemed low risk and showed a high risk of bias on only one domain or had an unclear risk of bias on at least half of the domains (4 out of 8) were considered medium risk. Studies with high risk of bias on more than one domain were considered high risk.

Meta-analysis

Using the program Comprehensive Meta-Analysis (Version 3.0; Borenstein et al. 2014), we conducted a meta-analysis on subjective and objective outcomes among studies that provided sufficient data for meta-analysis. One reviewer (CML) extracted data for meta-analysis and a second independent reviewer (RS) verified the results, with no discrepancies noted. Data were extracted for mean and standard deviation (SD) of the pretest and posttest values, mean and SD of change scores and sample size for each group, and t score or p value within groups. A pooled effect size was calculated for subjective and objective outcomes separately. Since the outcome variables were measured in different scales, the standardized mean difference (SMD) was used as an estimate of effect size. Subgroup analyses were also conducted to calculate effect sizes for meditation when compared to active controls versus inactive controls. For studies that included two control groups, we conducted two comparisons and divided the total N by 2 to avoid overestimation of the study. Given that very few studies included follow-up data, we focused the meta-analysis on immediate pre-post effects. We examined heterogeneity of the included studies based on the i-squared statistic and Q test to determine a fixed or random effects meta-analysis model according to the results (i-squared < 40% for fixed effects; Higgins and Green 2008). Publication bias was also assessed by funnel plot and the fail-safe N. We did not contact the authors to obtain missing data in order to prevent bias introduced by selective responding of authors.

Results

Literature Search

See Fig. 1 for details of our literature search and article selection process according to PRISMA guidelines (Moher et al. 2009). Our search yielded 479 results. After excluding duplicates (n = 282), unpublished manuscripts (n = 18), non-RCTs (n = 52), non-meditation interventions (n = 18), and studies that did not quantitatively measure prosocial outcomes (n = 43) or only measured self-focused compassion (n = 40), there were 26 studies that met our eligibility criteria.

Fig. 1
figure 1

Flowchart of article selection process

Full size image

Characteristics of Included Studies

Table 1 presents a summary of sample characteristics, meditation interventions, control interventions, and outcome measures across the 26 included studies (total N = 1714). Most studies (n = 22) were conducted in nonclinical adult populations using a primarily LKM or CM intervention (n = 10) or both mindfulness and LKM combined (n = 8). Studies did not tend to use protocolized interventions but rather incorporated similar elements to develop original protocols. Outcomes included various subjective and objective measures of empathy, compassion, and prosocial behaviors. All studies measured outcomes shortly after the end of the intervention; only four studies incorporated a longer-term follow-up (range = 8–52 weeks postintervention).

Table 1 Summary of characteristics of included studies
Full size table

Risk of Bias Assessment

Eleven studies showed a low risk of bias, 12 showed a medium risk, and 3 showed high risk (Tables 1 and 2). Four studies were classified as medium risk because risk was unclear on most domains, rather than because there were any high-risk domains. In general, studies showed lower risk of bias in terms of selective reporting of outcome measures (24 low risk), but higher risk of bias in terms of baseline imbalance (4 high risk), participant blinding, incomplete outcome data, and differential attrition (3 high risk each; see Fig. 2).

Table 2 Risk of bias for each study
Full size table
Fig. 2
figure 2

Risk of bias across all studies

Full size image

Synthesis of Results for Observable Outcomes

Table 3 presents the summary and results of each study. Most studies (11 out of 14; 79%) found support for improvements in observable outcomes following meditation as compared to the control intervention, with no clear difference in results by study quality. There were seven low-risk studies that measured observable prosocial outcomes and all of them reported improvements following meditation (Kang et al. 2014; Kemeny et al. 2012; Mascaro et al. 2013, 2015; Rosenberg et al. 2015; Schonert-Reichl et al. 2015; Weng et al. 2013). The majority of these studies used active control groups (n = 5; Kang et al. 2014; Mascaro et al. 2013; Schonert-Reichl et al. 2015; Weng et al. 2013, 2015). The remaining seven studies were medium risk, and four of these found support for improvements in observable prosocial outcomes as compared to active (Hutcherson et al. 2008; Logie and Frewen 2015) and wait-list controls (Condon et al. 2013; Flook et al. 2015). Two of the three studies that did not find significant effects were conducted in children (Poehlmann-Tynan et al. 2016; Velásquez et al. 2015). Two studies that reported observable improvements were the same studies that did not find support for subjective improvements (Kang et al. 2014; Rosenberg et al. 2015). Both of the studies that directly compared mindfulness and compassion meditation found no significant differences between them for improving prosocial outcomes (Condon et al. 2013; Logie and Frewen 2015). In studies with follow-up assessments, one found maintained gains at 5 months postintervention (Kemeny et al. 2012), and the other found no significant improvements postintervention or at follow-up (Poehlmann-Tynan et al. 2016).

Table 3 Summary of included studies
Full size table

A total of 11 studies provided sufficient data on observable outcomes for meta-analysis with 12 comparisons available (see Fig. 3). There was sufficient homogeneity among studies to conduct a fixed effects analysis (i-squared = .00; Q value = 1.82, p = .96). The effect size across these studies was .45 (p < .001; 95% CI = .28–.61). Results were similar for studies with active (SMD = .48, p < .001; 95% CI = .25–.72) and inactive control groups (SMD = .41, p < .001; 95% CI = .19–.63). Publication bias was not suspected based on the funnel plot (see Fig. 4) and the number of negative studies needed to make the results nonsignificant (N = 76, p = .92).

Fig. 3
figure 3

Effects of meditation on objective and subjective outcomes. Note. SDM, standardized difference in means; CI, confidence interval; LL, lower limit; UL, upper limit

Full size image
Fig. 4
figure 4

Funnel plots for a objective and b subjective outcomes

Full size image

Synthesis of Results for Self-Reported Outcomes

The majority of studies (14 out of 19; 74%) found significant improvements in self-reported outcomes following meditation compared to the control intervention for at least one prosocial outcome (e.g., empathy, compassion, or prosocial behavior). These results did not appear to appreciably vary depending on the study’s level of risk of bias. There were seven low-risk studies and four found support for improvements in self-reported empathy or compassion as compared to wait-list (Jazaieri et al. 2013; Shapiro et al. 2010) and active control groups (Kang et al. 2015; Schonert-Reichl et al. 2015). Three low-risk studies did not find support for subjective improvements (Kang et al. 2014; Keefe 1979; Rosenberg et al. 2015). There were nine medium-risk studies and eight found support for subjective improvements (Ashar et al. 2016; Asuero et al. 2014; He et al. 2015; Hutcherson et al. 2008; Logie and Frewen 2015; Oman et al. 2010; Shapiro et al. 1998; Taylor et al. 2015). Three studies were classified as high risk and two of these found improvements (Kok et al. 2013; Wallmark et al. 2013). Both studies that included a long-term follow-up found that improvements were maintained over time (Oman et al. 2010; Shapiro et al. 2010).

A total of 18 studies provided sufficient data on self-reported outcomes for meta-analysis. These studies allowed for 19 comparisons because one study used two control groups. Results indicated sufficient homogeneity to conduct a fixed effects meta-analysis (i-squared = .00; Q value = 3.94, p = .49). The effect size for subjective outcomes across these studies was .40 (95% CI = .28–.52, p < .001). The results were similar across studies that used active (SMD = .43, p < .001; 95% CI = .21–.65) and inactive control groups (SMD = .39, p < .001; 95% CI = .24–.53). Publication bias was not suspected based on the funnel plot and because the number of studies needed to make the results nonsignificant was 165 (p = .55).

Synthesis of Results for Potential Mechanisms

Fourteen studies reported results for potential mediators of effects of meditation on prosocial outcomes. Six of these studies conducted formal mediation analyses (Ashar et al. 2016; Hutcherson et al. 2008; Kang et al. 2014; Kok et al. 2013; Oman et al. 2010; Shapiro et al. 1998). Formal mediation results revealed that increased social and emotional connectedness mediated the effects of compassion meditation and charitable donations (Ashar et al. 2016), increased positive affect mediated the effect of LKM on explicit bias toward marginalized groups (Hutcherson et al. 2008), decreased stress mediated the effect of LKM on bias (Kang et al. 2014), and greater home practice and decreased stress mediated the effect of meditation on compassion (Oman et al. 2010). Kok et al. (2013) tested more complex structural models and found that loving kindness meditation led to improvements in positive emotions, which led to improvements in social connectedness, which led to improvements in vagal tone. Shapiro et al. (1998) found that greater meditation compliance led to decreased anxiety, which led to greater empathy.

Eight studies did not conduct formal mediation analyses but explored correlations between changes in prosocial outcomes and changes in other variables that suggest potential mechanisms of action (Jazaieri et al. 2013; Keefe 1979; Kemeny et al. 2012; Mascaro et al. 2013; Rosenberg et al. 2015; Velásquez et al. 2015; Wallmark et al. 2013; Weng et al. 2013). Almost all (7 out of 8) examined the relationship between amount of home practice/meditation adherence and prosocial outcomes: five found that greater meditation practice was correlated with greater prosocial outcomes (Jazaieri et al. 2013; Keefe 1979; Rosenberg et al. 2015; Velásquez et al. 2015; Wallmark et al. 2013) and two found no significant correlation (Kemeny et al. 2012; Mascaro et al. 2013). One study also found that increases in mindfulness and self-compassion, and decreases in stress, were significantly correlated with increases in empathy (Wallmark et al. 2013). Two studies used fMRI to explore correlations between prosocial outcomes and changes in neural function (Mascaro et al. 2013; Weng et al. 2013). Mascaro et al. (2013) found that improvements in empathy were correlated with increased activity in the inferior frontal gyrus (IFG) and dorsomedial prefrontal cortex (dmPFC). Weng et al. (2015) found that greater prosocial behavior (charitable donations) were correlated with changes in the inferior parietal cortex and dorsolateral prefrontal cortex.

Discussion

The results of the current systematic review support the efficacy of meditation-based interventions for increasing empathy, compassion, and prosocial behaviors. Meditation interventions showed significantly greater improvements in at least one prosocial outcome as compared to control groups in 22 out of the 26 included RCTs (85%). Meta-analysis results indicated that meditation training had a small-medium and significant effect on both subjective and objective prosocial outcomes, which was similar across studies with active and inactive control groups if not slightly higher among those with active controls. Many studies were low risk, with only three studies showing a high risk of bias and there were no clear differences in outcomes based on risk of bias. Effects for observable outcomes (e.g., real-world helping behavior, facial expressions) were somewhat stronger and more consistent than results for self-reported outcomes, though both showed significant improvements in the meta-analysis.

Results of several studies suggest potential mechanisms by which meditation can improve prosocial outcomes. Potential emotional mechanisms include an increased sense of social-emotional connectedness with others (Ashar et al. 2016; Kok et al. 2013), increased positive affect (Hutcherson et al. 2008; Kok et al. 2013), decreased stress and negative affect (Kang et al. 2014; Oman et al. 2010; Shapiro et al. 1998), and greater trait mindfulness and self-compassion (Wallmark et al. 2013). Some studies directly tested self-focused emotional mechanisms as mediators of meditation training on prosocial outcomes and found significant indirect effects, suggesting that meditation leads to improvements in individuals own socioemotional functioning and, thereby, improvements in prosocial outcomes (Ashar et al. 2016; Hutcherson et al. 2008; Kang et al. 2014; Kok et al. 2013; Shapiro et al. 1998). Consistent with the larger literature demonstrating that meditation interventions improve self-focused emotions (Hofmann et al. 2010; Kirby 2017), these results suggest that one way meditation practice can lead to improvements in prosocial emotions is by improving individuals’ own socioemotional well-being. These mechanisms are also consistent with research demonstrating that mindfulness-based interventions increase trait mindfulness (Quaglia et al. 2016), as trait mindfulness is likely to promote real-time awareness of others’ suffering and thus greater opportunities for prosocial action (Bibeau et al. 2016). Amount of meditation practice may play a role in a dose-response relationship, with reports of greater practice associated with greater improvements (Jazaieri et al. 2013; Keefe 1979; Oman et al. 2010; Rosenberg et al. 2015; Shapiro et al. 1998; Velásquez et al. 2015; Wallmark et al. 2013). However, it is possible that some studies did not find a relationship between home practice and outcomes and did not report these nonsignificant findings.

This synthesis has also identified potential physiological and neural mechanisms underlying these effects. Many meditation practices elicit physiological processes associated with the relaxation response (i.e., parasympathetic dominance), which is the physiological counter to the stress response (i.e., sympathetic dominance; Benson 1997). Regular elicitation of the relaxation response is associated with reduced stress and negative emotions (Esch et al. 2003) and is thought to play a role in improving prosocial emotions (Kirby 2017). In the current review, meditation was indeed associated with improvements in vagal tone (Kok et al. 2013 ). Meditation was also associated with altered activation in areas of the prefrontal cortex (Mascaro et al. 2015; Weng et al. 2013). These findings are similar to previous studies of meditation for general health outcomes (Marchand 2014; Pace et al. 2009) and non-RCTs of meditation for prosocial outcomes (Klimecki et al. 2012, 2014; Leiberg et al. 2011) and further support a neural and physiological basis for meditation’s effects on prosocial outcomes specifically.

Although not emphasized in most of the studies included in the current review, meditation-relaxation physiology may be associated with improved prosocial outcomes through oxytocin-mediated improvements in attachment style. The same physiological processes that characterize the relaxation response have been shown to occur in the context of secure attachment and mother-child dyads, which provide a foundation for compassion (Fricchione 2011; Hill-Soderlund et al. 2008; Mikulincer et al. 2005; Oosterman et al. 2010). Oxytocin plays a role in both relaxation and secure attachment physiology and is also associated with greater prosocial behaviors (e.g., improved face expression recognition, enhanced encoding of positive social memories; Isgett et al. 2016; Mascaro et al. 2015; Strathearn et al. 2009). If meditation stimulates oxytocin receptors and mimics the physiology of secure attachment, then it is reasonable and researchable to hypothesize that meditative approaches will enhance prosocial behaviors (Kim et al. 2014; Rilling 2009; Strathearn et al. 2009). Only one study included in the current review directly addressed the potential role of oxytocin, by using a placebo oxytocin control group; results indicated greater improvement in subjective but not objective prosocial outcomes among CM participants than oxytocin placebo participants. Recent theories highlight that the role of oxytocin in social behavior is complex and not necessarily prosocial, depending on individual difference characteristics (e.g., gender, psychopathology; Shamay-Tsoory and Abu-Akel 2016). Future research should explore whether oxytocin is another physiological mechanism by which meditation leads to enhanced prosocial benefits.

Another potential mechanism of action that was not emphasized in the current systematic review and has not been explored in any of the studies included here involves emotional tolerance and regulation. Beyond reductions in level of emotional problems, improvements in the way individuals withstand or respond to negative affect might also play a role (Mascaro et al. 2015). Theoretical conceptualizations of compassion emphasize that individuals must be able to tolerate the distress they feel in response to another’s suffering in order to effectively enact helping behaviors (Strauss et al. 2016). Distress tolerance, an individual difference variable defined as the ability to withstand negative affective states (Simons and Gaher 2005), is a well-established risk factor for emotional disorders that influences emotion regulation strategies (i.e., low levels of distress tolerance motivate maladaptive avoidance). Meditation interventions, particularly mindfulness meditation, have been shown to significantly increase distress tolerance and improve emotion regulation (Chambers et al. 2009; Lotan et al. 2013), and emotion regulation is thought to play a role in the effects of LKM on prosocial outcomes (Mascaro et al. 2015). Thus, meditation might also improve compassion and other prosocial outcomes by improving the way individuals tolerate and respond to distress, in addition to decreasing the amount of distress an individual experiences. Future research should directly test these potential mechanisms.

The current findings are supported by the relatively strong design and low risk of bias across many RCTs, the homogeneity of studies included in the meta-analysis, and evidence for lack of publication bias. Many studies used active control groups and objective behavioral outcomes, and meta-analysis results were similar across type of control group and outcome measure. However, samples were all nonclinical and primarily female and White, and half did not report the racial composition of the sample. Greater sample diversity is needed and future studies should describe the full demographic characteristics of the sample. Describing the details of the randomization procedure and concealment and maintaining participant blinding (e.g., concealing the true intent of the study, using active matched control groups) could also further improve the methodological rigor of future studies.

Nonetheless, this review highlights several directions for future research. First, research on more clinically and demographically diverse samples is needed to enhance generalizability. Second, although the meta-analysis indicated homogeneity among studies, there was variability among the meditation interventions. Future studies may consider using manualized protocols or conduct dismantling studies to establish optimal intervention dose and content. In addition, research should examine a wider range of meditation types and formats, such as movement-based meditations and individual (rather than group) in-person interventions. This research should also include comparative efficacy trials that directly compare different types of meditation and other evidence-based interventions that improve emotional problems (i.e., traditional cognitive-behavioral therapy). In the current review, most studies incorporated LKM, which is a relatively newer research area as compared to mindfulness meditation, and found significant prosocial benefits. Moreover, both of the studies that compared LKM to mindfulness did not find significant differences in prosociality, though Logie and Frewen (2015) found a greater effect of LKM on reducing self-positivity bias as compared to mindfulness meditation, and other previous studies have found some differences in emotional outcomes across meditation types (Zeng et al. 2015). Future studies should also incorporate longer-term follow-ups. These findings provide further support for continued research on LKM and the need for comparative efficacy work.

It is also worth noting that some research suggests empathy and compassion may have different utility for the person giving versus receiving help, particularly when empathizing with another’s suffering. Empathy (affect-sharing) may increase personal distress and reduce prosocial behavior, while compassion (affect-sharing with motivation to help) may strengthen personal resources and promote positive outcomes (e.g., Klimecki et al. 2014; Singer and Klimecki 2014). It is possible that these differential effects could vary depending on the individual’s own general ability to tolerate emotional distress. We included empathy to be comprehensive in our review of prosocial outcomes, but further research on the differential effects of empathy and compassion is warranted.

The current findings also have implications for clinical practice and meditation teachers in nonclinical settings. Clinicians and meditation teachers should be aware that meditation interventions (e.g., MBSR, MBCT) could provide additional benefits beyond reduced emotional distress. Clinicians might select meditation-based protocols for patients who are specifically interested in increasing empathy and compassion (e.g., parents, healthcare providers), or consider incorporating meditation training into other evidence-based interventions to maximize improvements for individuals experiencing interpersonal problems. Results suggest that integrating meditation training into other evidence-based interventions may be feasible, as even two weeks of 20 minutes daily practice via mobile phone applications have shown significant prosocial benefits. Meditation teachers in nonclinical settings should be aware that there is a scientific evidence base to support the broader prosocial benefits of individual meditation training, teach meditation with these benefits in mind, and consider discussing these potential benefits with students.

Limitations

In the current systematic review, limitations include heterogeneity in the interventions and an inability to include non-English studies, which may have biased the results and limits generalizability. Nonetheless, these results advance the scientific understanding of meditation for health outcomes and suggest that meditation training is a promising way to increase individual-level prosocial outcomes. Improving these prosocial outcomes has the potential to promote important societal changes needed today. Further research using more diverse samples and meditation practices is warranted.