Which Meditation Technique for Whom? An Experimental Single-Case Study Comparing Concentrative, Humming, Observing-Thoughts, and Walking Meditation

Fifty-four participants were recruited through different university mailing lists and handbills distributed on the university premises of the Chemnitz University of Technology. We aimed to recruit 10–12 participants per condition to keep experimental group sizes at an optimal level. Individuals were excluded from participation if they (1) had preexisting psychiatric conditions or acute psychological issues, (2) had a regular meditation practice during the last 6 months, (3) were younger than 18 years, or (4) did not have daily access to web-enabled devices. Four participants had to be excluded and seven withdrew from participation prior to the study (Fig. 1). Forty-four participants were initially randomized. One participant dropped out immediately after randomization and prior to data collection, and one participant dropped out during the first weeks of the study, both due to scheduling conflicts. Another participant switched her time slot after randomization without knowing which treatment took place when. The final sample consisted of 42 participants (81% women, M_age = 24.64), mainly students and some employees (50% university staff, 50% other). Sociodemographic data according to condition can be found in Table 1.

During an information event that took place before randomization, the participants were fully informed about the nature of the study. They could not choose their treatment, but could indicate a preferred time frame (mid-day or afternoon). At this point, neither researchers nor participants knew which treatment took place at what time. Participants only learned about their allocated meditation technique during their first group session. For their participation in the full study, they received 50 € or course credit (or a combination of these).

During baseline, participants engaged in their usual daily activities. The treatment conditions were concentrative meditation (CON), humming meditation (HUM), observing-thoughts meditation (OBS), and walking meditation (WALK). We randomized participants across these four conditions and across three baseline lengths (14, 21, or 28 days) into 12 subgroups using a computer-based randomizer. The individual treatment length differed depending on participants’ baseline length—participants with a 14-day baseline completed 8 weeks of treatment, with 21-day 7 weeks, and with 28-day 6 weeks. This was done to evaluate potential effects of treatment length. Participants received daily online questionnaires throughout their entire baseline and treatment phases. Sociodemographic variables and individual factors were collected prior to the beginning of the baseline phase. The study design is displayed in Fig. S1 in the Supplementary Materials.

All measures were taken online using SoSci Survey (Leiner, 2019). Data were collected between 22 January 2020 and 31 March 2020. Hence, the final weeks of the study fell within the first global COVID-19 outbreak with the first national lockdown in Germany starting on 22 March. The COVID-19 situation did not impact the measurements in this study as they were carried out consistently online and participants continued responding throughout the entire study period. We conducted a follow-up in the beginning of June, when most governmental restrictions had been loosened again. We collected data for all dependent variables continuously throughout the study period. To reduce response burden, we obtained data for some questionnaires once or twice per week and collected only well-being data on a daily basis. For the multi-layered construct of body awareness, we employed a newly developed questionnaire that we successfully implemented in earlier studies (Matko et al., 2022b) and that is currently being validated (Bergt & Matko, 2023). To suit the needs of twice weekly measurements, we extracted the 11 most suited items from four established questionnaires (Baer et al., 2006; Mehling et al., 2012; Miller et al., 1981; Shields et al., 1989). We measured emotion regulation weekly with a shortened version of the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2003), in which we extracted the two items with the highest factor loadings per subscale, resulting in altogether 12 items. To increase readability, we inverted participants’s scores on this measure so that higher scores represent higher emotion regulation skills.

Decentering, which implies having a distanced perspective on one’s internal experiences, was measured weekly with the 11-item Experiences Questionnaire-Decentering Scale (EQ-D; Fresco et al., 2007). Spontaneous mind-wandering reflects behavior when the mind wanders inadvertently or uncontrolled. It was assessed twice weekly using the 4-item Mind-Wandering: Spontaneous scale (MD-S; Carriere et al., 2013). Participants’ well-being was measured daily with the 5-item World Health Organization Well-being Index (World Health Organization, 1998). The WHO-5 is a psychometrically sound self-report measure of well-being with high internal consistency and high convergent validity (Brähler et al., 2007). We adapted the time frame of this measure to “the last 24 h.” Well-being was rated on a 6-point Likert scale, mind-wandering on 7-point scale, and all other abovementioned questionnaires on a 5-point scale.

During pretest, we assessed several possible moderators of treatment outcome. We utilized the 10-item Big Five Inventory (BFI-10; Rammstedt & John, 2007) to capture the Big Five personality traits: extraversion, neuroticism, openness to experience, conscientiousness, and agreeableness. It is considered a concise and sufficiently reliable and valid instrument. Trait absorption describes a person’s disposition to become fully engaged (“absorbed”) in experiences and was related to meditation outcomes (Hölzel & Ott, 2006; Matko et al., 2022a). We measured it using the 34-item modified German version of the Tellegen Absorption Scale (Ritz & Dahme, 1995). On both scales, items were rated on a 5-point Likert scale.

In addition, we assessed participants’ motivation to meditate. We drew the six main motivational themes from Sedlmeier & Theumer, (2020)—coping with stress, mental improvement/developing skills, reducing physical or mental suffering, relaxation, seeking spiritual experience, and self-exploration—and asked participants to rate the importance of each theme on a 5-point scale at pretest, posttest, and follow-up. We also provided them with a free-text item to describe any other motivational reasons.

With the beginning of the treatment phase, we asked participants to indicate how long they meditated each day and how they subjectively experienced their meditation. Among others, we assessed the perceived difficulty of each meditation session using a 5-point polar profile. Detailed analyses on these variables and adverse events related to the meditation practice will be reported in another publication (Matko & Sedlmeier, unpublished). Furthermore, participants had the opportunity to describe any special events that occurred throughout their day in a free text item in the daily questionnaire to provide context for their responses. Finally, we collected qualitative data on how participants experienced their course and meditation practice during the weekly meetings (minutes) and retrospectively at posttest and follow-up (free text items).

Single-case data can be analyzed in multiple ways (Machalicek & Horner, 2018; Shadish, 2014), the most common being visual inspection of plots of the dependent variable over time (Lane & Gast, 2014). Often, multiple analysis strategies are chosen to increase the reliability and generalizability of findings. We analyzed our data on an individual level using visual analysis, effect size estimation and multivariate plots, and on a comparative level by performing correlational and meta-analyses. We chose a primarily exploratory and descriptive mixed-methods approach.

The majority of our participants conscientiously practiced their meditation technique for 15–20 min each day. The mean practice duration varied significantly between conditions: M_CON = 16.7, SD = 7.0; M_HUM = 14.0, SD = 8.0; M_OBS = 17.6, SD = 6.6; M_WALK = 14.4, SD = 8.3; F(3, 1807) = 25.87, p < 0.001. Thus, participants in the OBS and CON conditions meditated longer than participants in the HUM or WALK conditions, t₍₁₅₇₀₎ =  − 8.40, p < 0.001. Figure S2 depicts the mean trajectory of reported meditation duration in each condition (Supplementary Material). We plot only the first 6 weeks of practice of each participant to make time intervals comparable. Meditation practice duration was fairly constant in each condition, with CON participants slightly increasing their practice duration and WALK and HUM participants slightly reducing it.

Course adherence was moderate to high. On average, participants attended 74.6% of all group sessions. There were no significant differences in course attendance between conditions. However, course attendance was higher in shorter interventions, F(1, 10) = 19.16, p = 0.001, with a mean attendance of 80.9%, 76.5%, and 66.4% for the 6-, 7-, and 8-week courses, respectively. Participants were generally satisfied with their course and their meditation technique (the maximum rating being 5)—CON: Mdn = 3.83; HUM: Mdn = 3.67; OBS: Mdn = 4.17; WALK: Mdn = 3.92—with no significant differences in course satisfaction between conditions.

Figure 2 illustrates the development of body awareness scores of each participant over time. Most participants’ body awareness fluctuated to some extent with only some participants exhibiting stronger fluctuation. There were observable interindividual differences in the general level and slope of the curves.

The slopes during the baseline phase varied across participants with 15 participants showing stable body awareness during baseline, 14 exhibiting an increase, and 12 a decrease. Interestingly, around two thirds of the participants showed a noticeable jump with the beginning of the treatment phase—in either direction, with some participants suddenly reporting enhanced body awareness while others reported diminished body awareness. In either case, and for most participants, body awareness increased linearly over the course of the treatment. The variability was comparably low in both phases. Comparing the two phases, most participants improved their body awareness during the treatment. Seven participants showed no observable changes and only one participant per condition experienced a deterioration of body awareness.

Next, we calculated Tau-U for each participant and conducted a meta-analysis, which yielded a mean Tau-U = 0.27, 95% CI [0.17, 0.38], indicating a significant moderate improvement of body awareness. Eighteen effects yielded significant results with moderate to large effect sizes in the expected direction. However, two effects were significantly negative—Participants 23 (OBS) and 34 (WALK), the latter representing a noticeable outlier. In addition, a substantial amount of variation was not accounted for in the full model (I² = 71.4%, τ = 0.28 – τ is the standard deviation of the mean Tau-U). However, when we excluded the outlier, heterogeneity decreased substantially (I² = 61.9%, τ = 0.23) and the overall effect increased to Tau-U = 0.30 [0.21, 0.39]. Hence, we continued our analyses excluding this participant. Figure 3 depicts the resulting separate meta-analyses for each condition as well as the overall effect.

When we included subgroup as a moderator in the meta-analysis, it did not significantly differentiate between the subgroups, \({\chi }_{3}^{2}\) = 1.34, p = 0.720. This, together with the meta-analytic results shown in Fig. 3, indicates that all four meditation techniques had a similarly positive effect on body awareness. WALK had the lowest heterogeneity suggesting that this condition had the most reliable effect on body awareness.

For the remaining variables, all figures displaying the continuous measurements and full meta-analytic forest plots can be found in the Supplementary Material S3–S10. The summary versions of the forest plots are displayed in Fig. 4 a–d.

The visual analysis indicated that decentering did not fluctuate much (Supplementary Fig. S3). Only few participants exhibited a somewhat higher variation. Thirteen baseline trends could not be determined as these participants had less than three measurement points in the baseline. The analysis of trends indicated that this skill was rather stable during baseline, but exhibited a clear upward trend for the majority of participants during the treatment phase. Six participants showed a sudden decrease in decentering in the last 2 weeks of the study, coinciding with the onset of the pandemic. For some participants, decentering remained stable throughout the treatment and only two participants showed a slight downward trend. Comparing both phases, most participants increased their decentering skills over time, while only few participants showed no change or a decrease. This positive change was mostly gradually, indicating that decentering improved with continued meditation practice. Six participants experienced a sudden drop of decentering with the beginning of the treatment, suggesting a reinterpretation of the scale after the first meditation. For all of these participants, decentering gradually improved over the course of the treatment reaching or exceeding reported baseline levels.

The meta-analysis suggested a moderate improvement of decentering, Tau-U = 0.35 [0.24, 0.46], (Fig. 4a and Supplementary Fig. S4). Fourteen effects yielded significant results with moderate to very large effect sizes. There was no significant moderation effect for subgroups indicating that all four techniques had a similar effect on decentering. However, the distribution of effects differed slightly across the four subgroups. CON and OBS yielded more consistently positive effects. The effects of HUM and WALK were not significant and more heterogeneous.

Visually, emotion regulation fluctuated to some extent in both phases (Supplementary Fig. S5). Fourteen baseline trends could not be determined, but this skill generally appeared to be rather stable. In the treatment phase, some participants showed a clear upward trend, others remained more or less stable, and still others fluctuated rather inconsistently. Only three participants exhibited a downward trend. Comparing both phases, the majority of participants improved their emotion regulation skills; some participants showed no change and only few experienced a deterioration over time.

The meta-analysis indicated a moderate improvement, Tau-U = 0.31 [0.19, 0.44] (Fig. 4b and Supplementary Fig. S6). Fourteen effects yielded significant results with moderate to very large effect sizes. All meditation techniques had a similarly positive effect and the distribution of effects was comparable across the four subgroups. All subgroup effects were significant, except the effect of the HUM condition. The WALK condition had the lowest heterogeneity indicating a more consistent effect.

Mind-wandering should decrease over the course of the treatment. Thus, a downward trend represents an effect in the expected direction. The visual analysis indicated that variability in mind-wandering scores was similar in both phases and overall, relatively high (Supplementary Fig. S7). It differed strongly across participants with some participants displaying very high and others very little variability. Two participants reported almost no mind-wandering from the beginning and had, thus, little room for improvement. The trends in the baseline and treatment phases varied. In the treatment phase, the majority of participants exhibited a clear downward trend, while for some, mind-wandering remained fairly constant or increased over time. However, quite a few participants reported increased mind-wandering during the last week of the study, coinciding with the pandemic.

Comparing both phases, the effects were inconsistent. Interestingly, several participants experienced an upward jump in mind-wandering with the beginning of the treatment, which for most continuously decreased over the following weeks to baseline or below-baseline levels. This corresponds with qualitative reports of participants who stated being surprised at how “crowded” their minds were during their first meditation and how they became quieter over time. Only few participants experienced an immediate downward jump. For others, the transition between baseline and treatment was smoother with mind-wandering levels either declining or staying the same. Only four participants had markedly higher levels of mind-wandering compared to baseline.

The meta-analysis on Tau-U effect sizes was similarly inconsistent, although it indicated a small improvement, Tau-U =  − 0.14 [− 0.23, − 0.05] (Fig. 4c and Supplementary Fig. S8). Eleven effects yielded significant results with moderate to large effect sizes. There were no significant differences between the subgroups. However, only the OBS condition yielded a significant subgroup effect, and the OBS and CON conditions had the most participants with significant effects. The WALK condition had the highest heterogeneity and an effect close to zero, suggesting this condition might be the least effective in reducing mind-wandering.

Daily well-being fluctuated considerably in both phases, for some participants stronger than for others (Supplementary Fig. S9). Most participants showed no trend during baseline, but a slight linear increase in the treatment phase. For other participants, the level of well-being remained more or less the same during the treatment and/or showed a wave-like distribution with no clear trend. Only two participants exhibited a slight downward trend during the treatment. Interestingly, the well-being of some participants tended to fluctuate less towards the end of the treatment suggesting a stabilization. Comparing both phases, the treatment appeared to improve well-being to some extent, but not substantially. Most transitions were smooth between baseline and treatment, apart from some jumps in either direction. Only three participants experienced a small decline of well-being over time.

The meta-analysis suggested a small improvement, Tau-U = 0.17 [0.12, 0.23] (Fig. 4d and Supplementary Fig. S10). Twenty effects yielded significant results with small to moderate effect sizes. All subgroup effects were significant and of a similar size. However, the distribution of effects and the heterogeneity differed between groups. OBS had the largest effect, but also the highest heterogeneity. Heterogeneity was similarly high for CON, but considerably lower for HUM and WALK, indicating these conditions had a more reliable effect on well-being.

As our experimental groups differed in the distribution of age, gender, and occupation in the different groups, we performed meta-regression analyses on each dependent variable. We entered the variables age, gender, and occupation as predictors into each meta-analytic model. All resulting regression coefficients and their corresponding confidence intervals are displayed in Supplementary Table S11. There were no significant meta-analytic regression coefficients for any variable. On an alpha-level of 10%, we found a very small effect for age in well-being (b = − 0.02 [− 0.03, 0.00]), and a medium effect of gender in emotion regulation (b = − 0.32 [− 0.66, 0.02]), indicating that women might have increased their emotion regulation skills more than men.

Only very few participants exhibited negative effects on any of the dependent variables. We found one outlier for body awareness, but could not find any explanation for this effect in her qualitative data (Case 34, WALK). In addition, she improved in most other variables. The same was true for most other cases with negative effects. Most were small in size and limited to one specific variable. To get a more systematic impression of different response patterns between participants, we plotted the individual responses to the five different variables in multivariate star plots (Fig. 5). Negative effects are represented by non-existent or very short lines and the four conditions are color-coded.

Figure 5 reveals interesting response patterns. Four participants benefited equally regarding all five dependent variables (3, 4, 25, 29). Two others showed a similar pattern, albeit on a smaller scale (5, 33). All of these participants had learned either CON or OBS meditation. Many participants showed considerable improvements, but not on all variables (6, 7, 14, 17, 24, 30, 32, 37, 38, 39). Some participants benefited mainly regarding body awareness, decentering, and emotion regulation (9, 18, 40, 41). Furthermore, we identified groups of participants who improved mostly on two variables—either body awareness and decentering (8, 10, 21, 27, 28, 42), mind-wandering and emotion regulation (13, 22, 34), or body awareness and well-being (2, 26, 36). Some participants exhibited only small improvements on some variables (11, 12, 16, 19, 31). Finally, four participants, one per condition, did not benefit from the treatment regarding a majority of variables (1, 15, 23, 35). We will now have a closer look at the results of these four participants.

Participant 1 (CON) did not improve in any of the variables. She generally found meditating quite hard, but stated that the group meetings had helped and supported her. Although she meditated regularly, her practice duration fluctuated a lot. At follow-up, she stated that she was still meditating regularly, but less often and with a different technique. Moreover, she reported that many things did not stress her out as much as before the course. Participant 15 (HUM) improved only regarding body awareness. He meditated conscientiously, although he repeatedly stated having difficulties as he experienced boredom, restlessness, or troubling thoughts. At follow-up, he described how meditating did not necessarily relax or calm him. Rather, it let suppressed things come to the surface and forced him to confront these, which he did not always succeed in.

Participant 23 (OBS) did not benefit regarding any of the variables. She reported a lot of stress during the study as she was finishing her master’s thesis and moving to a new apartment. She conscientiously meditated throughout the treatment and reported using meditation to calm down and destress. During the weekly meetings, she described different experiences during meditation—her legs feeling detached from her body, a flood of pictures, or that she had to discontinue meditation when it became too much for her. At follow-up, she stated that the course was a good and valuable experience that helped her to realize the importance of taking time to rest and reflect. Participant 35 (WALK) enhanced her body awareness substantially, but showed no improvement in any of the other variables. She meditated regularly at the beginning of the treatment, but discontinued her practice in the final 3 weeks. She reported relationship problems and moving to another house during this time. At posttest, she stated that meditating helped her to calm down and calm strong emotions on some, but not all days.

We had included the subgroup variable in all five meta-analyses as a moderator, but the respective tests did not reach significance in any case, indicating that the meditation technique used did not make a general difference. All four meditation techniques showed a similar pattern in their effects on the five dependent variables. The only condition that stands out in this respect is the WALK condition. Compared to the other conditions, it had the lowest effect on decentering and mind-wandering but the highest and most consistent effect on body awareness and emotion regulation. To simplify the following correlational analyses, we decided to calculate an aggregated treatment effect. For this purpose, we first reversed mind-wandering scores and then calculated the mean for the effect sizes of all five dependent variables for each participant (see Supplementary Table S12 for a full table of aggregated effects).

Next, we evaluated whether personality-related and motivational factors were correlated with the total treatment outcome in each of the four meditation groups. Supplementary Table S13 displays the resulting correlation coefficients and corresponding p-values. Moreover, we tested each pairwise correlation coefficient for significance with a function developed by Moore (2022). Due to the exploratory nature of our research, we did not correct for multiple comparisons. The resulting p-values for each comparison are displayed in Supplementary Tables S14 and S15. Figure 6 displays the correlations per condition for each personality trait in a comparative bar chart. Two thin horizontal lines were drawn at r =  ± 0.30 to delineate effects that are considered moderate to large. Asterisks were added to indicate significant correlations (within bars) and comparisons (between bars).

Figure 6 depicts a number of moderate to very large correlations within the different conditions. Agreeableness showed a moderate to large correlation with total treatment outcome in the CON and HUM condition, but a very large negative correlation in the WALK condition. Hence, participants high in agreeableness benefitted significantly less from walking meditation. Extraverted participants, on the other hand, obtained greater treatment gains from the two less known meditation techniques (HUM and particularly WALK), while the opposite applied to the two more traditional techniques (CON and OBS). In addition, participants high in neuroticism benefitted more from the two traditional techniques. While the correlation coefficients for these two conditions were significant, the differences between conditions were not. Lower openness to experiences was associated with a more positive treatment outcome in the HUM condition, but differences were not significant.

Likewise, Fig. 7 displays a number of large correlations with the motivational factors. Participants in the CON condition benefitted more if they had a high motivation to develop certain skills and a low motivation to seek spiritual experience. Participants practicing HUM showed greater treatment gains if they had a low motivation to improve skills or a low spiritual motivation. In contrast, participants practicing OBS benefitted more if they sought spiritual experiences and less if they aimed for relaxation or reduced stress. Finally, participants in the WALK condition experienced significantly better treatment effects if they were motivated to reduce stress, but benefitted less if they wanted to reduce suffering or sought spiritual experiences.

One aim of this study was to evaluate mechanisms of meditation and trajectories of change. The 6–8-week training particularly affected body awareness, decentering, and emotion regulation skills. This conforms to two theoretical proposals that deemed these skills central to mindfulness meditation (Bennett et al., 2021; Hölzel et al., 2011). Our findings indicate that these proposals (partially) extend to other forms of meditation, as well.

For the majority of participants, we observed a linear increase in these three skills over time. This indicates that continuous practice leads to steady improvements, at least during the first weeks of meditation practice. These trends were mostly independent of the actual treatment length and daily practice time (Matko & Sedlmeier, unpublished). However, as most participants conscientiously adhered to their practice, there might have been too little variance to evaluate the latter aspect. In a methodologically strong dismantling study, Lindsay et al. (2018) found that learning acceptance skills was a critical ingredient to boost the effectiveness of mindfulness interventions. As all of our treatments encouraged participants to approach their experiences with acceptance, this might have led to similar learning experiences.

For some variables and some participants, we observed sudden jumps between baseline and treatment phase, which could be indicative of expectancy effects. However, these jumps occurred in both directions. This suggests that some participants might have suddenly realized, for example, how crowded their mind was, how little they knew about their body and emotions, and what the items on the decentering scale actually meant. Our study design enabled us to monitor these changes in high temporal resolution. For instance, some participants reported decreased decentering or increased mind-wandering during the last 2 weeks of the treatment, coinciding with the onset of the COVID-19 pandemic.

Although we observed substantial interindividual differences, we found relatively consistent effects for body awareness, decentering, and emotion regulation. As these three skills are theoretically relevant, we assume they are, indeed, key mechanisms of meditation, at least at the beginning phase. Enhancing these skills substantiates the usefulness of diverse meditation techniques in promoting mental health (van Agteren et al., 2021).

Our design enabled us to examine group differences in a detailed way. With respect to the different dependent variables, response patterns were similar in all four meditation groups, with some participants benefitting a lot and others not at all. Although our design was not explicitly aimed at establishing group differences and the moderator tests did not reach statistical significance, we did observe some interesting commonalities and differences between the various meditation techniques.

All techniques led to moderate and (mostly) linear increases of body awareness and decentering. WALK and HUM had the most reliable effects on body awareness, which might be due to the increased attention to the moving or humming body. OBS and CON enhanced decentering more reliably. These two techniques are less active than HUM and WALK (see classification system by Matko & Sedlmeier, 2019), which might have encouraged participants to engage with and detach from their thoughts more. All techniques moderately enhanced emotion regulation skills; WALK was the most reliable in this respect. Its focus was the farthest away from the “emotional centers” of the body, possibly facilitating emotion regulation. This result might be seen as consistent with the findings of Singh et al. (2011, 2019) in their studies examining the effects of a meditation “on the soles of the feet.”

Mind-wandering and well-being fluctuated more than the other variables. Both variables improved to a small extent, but effects on mind-wandering were less consistent. OBS had the most reliable effect on mind-wandering (and the strongest focus on observing the wandering mind), while WALK had an effect close to zero. In contrast, WALK and HUM had the most reliable effects on well-being. A possible explanation might be that these two more active techniques provided a stronger contrast to participants’ (cognitively demanding) daily life, thereby enhancing their restorative effect (Kaplan, 2001). Likewise, an experience-sampling study indicated that mindful walking in nature significantly improved mindfulness and positive affect (Gotink et al., 2016). The effects of meditation on well-being might seem surprisingly small, but they actually correspond with previously reported findings (Goyal et al., 2014; Sedlmeier et al., 2012).

It is difficult to compare our results to previous studies as no other study has compared a similar set of meditation techniques. Looking at the available findings, many studies find commonalities between the compared meditation techniques, for example, regarding mindfulness skills (Feruglio et al., 2021), acceptance of stressors and reduced distress (Lohani et al., 2020), positive emotions (Fredrickson et al., 2017), or rumination, self-compassion, and well-being (Sauer-Zavala et al., 2013). At the same time, all of these studies report differential effects, too.

The ReSource project, a large 9-month trial consisting of three 3-month modules, compared the effects of different types of meditation on a range of variables (Singer & Engert, 2019). They observed similar differences in the subjective experience of breathing meditation, the body scan, loving-kindness meditation, and observing-thoughts meditation with respect to daily experience (Kok & Singer, 2017), perceived effort and likability (Lumma et al., 2015), and descriptions in micro-phenomenological interviews (Przyrembel & Singer, 2018). Breathing meditation and the body scan reliably increased interoceptive awareness (Bornemann et al., 2015). However, this increase was observed predominantly in aspects related to self-regulation rather than a basic awareness of body sensations—in contrast to our study that primarily measured the latter aspect. In a conclusive statement, Singer and Engert (2019) acknowledge the need for more granularity in research, that is, a more systematic investigation of the specific effects of different practice types and their subcomponents.

We observed a considerable variability between participants in their response to the different treatments and the different variables, pointing to substantial interindividual differences. Fittingly, in another manuscript (Matko & Sedlmeier, unpublished), we describe that the four conditions differed significantly in the subjective experience and adverse events they elicited. For example, participants meditated longer in the CON and OBS conditions, even though CON was perceived as the most difficult and least relaxing. OBS and HUM were perceived as easy, but elicited the most adverse or extraordinary events. In contrast, participants in the WALK condition had the least adverse events and were the most relaxed and awake during meditation. Several authors have pointed out that the subjective experiences and outcomes of meditation interventions depend strongly on individual factors (Hölzel et al., 2011; Schmidt, 2014; Tang & Braver, 2020), which we explored in this study.

We observed that participants high in extraversion benefitted more from the two less common techniques, while participants who scored high in neuroticism exhibited a tendency to having higher gains when they practiced one of the two “classic” meditation techniques, although the latter differences between conditions were not significant. Until recently, meditation techniques were mostly classified into focused attention (FA) and open-monitoring (OM) practices (Lutz et al., 2008). This classification has been called into question and some researchers proposed more inclusive and diverse classification systems (Dahl et al., 2015; Matko & Sedlmeier, 2019; Nash & Newberg, 2023). Nevertheless, FA and OM remain the two best-known classes of meditation practices, in general. In our study, two techniques correspond to these two types—concentrative meditation (FA) and observing-thoughts meditation (OM). Hence, we can regard these two techniques as more “classic” than the other two techniques, humming and walking, which are much less common in research and practice (Matko et al., 2021a). Interestingly, we found more consistently positive response patterns in the two classic techniques. This might either be indicative of a higher expectancy effect for more common techniques or underpin their suitability for beginners.

Our findings correspond with the characteristic conception of extraversion and neuroticism. Extraverted individuals tend to be active and enjoy stimulation (Hagan & Hausenblas, 2005), which humming and walking meditation are able to provide. The opposite applies to individuals high in neuroticism, who tend to avoid ambiguous experiences (Lommen et al., 2010) and might be less inclined to engage in “unusual” meditation techniques. It should be mentioned, however, that the type of concentrative meditation used in this study differs considerably from the types used in other studies, which mostly employed focusing on the breath (e.g., Ainsworth et al., 2013). In contrast, high levels of agreeableness were detrimental in walking meditation. This is in line with another study, where agreeableness predicted greater use of sitting and informal meditation in an MBSR program (Barkan et al., 2016).

Recently, some studies investigated the reasons why meditators start or continue meditation (Jiwani et al., 2022; Sedlmeier & Theumer, 2020). Surprisingly, there is almost no research on how different motivational components affect outcomes of meditation interventions. Hence, our study provides a promising first glimpse into this matter. Often, we observed beneficial effects if the motivational reason corresponded to the modality of or implicit assumptions regarding the meditation technique. Concentrative meditation was beneficial for those seeking to improve their (concentrative) skills, observing-thoughts meditation for those with a spiritual motivation, and walking meditation for those who sought better ways to cope with stress. Fittingly, the initial motivation to engage a specific meditation anchor (breath, image, phrase) was a major predictor of ultimate anchor preference (Anderson & Farb, 2018). Likewise, the belief of receiving mindfulness instead of sham meditation predicted increased pain threshold and tolerance, irrespective of the actual treatment participants received (Davies et al., 2022). Hence, also participants’ expectations, motivation, and preferences should be taken into consideration in meditation interventions.

This study was the first to systematically investigate four basic meditation techniques using a rigorous methodological design. It provides unique insights into the working mechanisms of meditation and the processes of change that take place when people begin to meditate. Although we did our best, we could not control all possibly influential factors. Participants were randomized across conditions, but we could not control the higher rates of employees and older young adults in the afternoon conditions compared to the mid-day conditions. However, our meta-regression analyses suggested our results were not significantly influenced by these differences. We recruited participants on university premises and tried to enable participation to all those interested in the study. This was another limitation—all of our participants had completed high school or higher education and were, thus, not representative of the general public. However, this is a problem commonly reported in positive psychology research (Hendriks et al., 2019). As our study was quite demanding in terms of data collection, we compensated participants for their participation. This effectively reduced the amounts of missing data and attracted participants who might not have been fully intrinsically motivated to participate in the study. Nonetheless, most participants (not all) reported high levels of intrinsic motivation and interest in meditation at the beginning of the study. However, recent research suggests that meditation interventions could only be effective for those who are interested in learning these skills (Montero-Marin et al., 2022).

For single-case research, our sample size of 42 participants and the number of dependent variables measured were exceptionally large. Therefore, our visual analysis was kept relatively simple and could certainly be executed in a more detailed manner. Moreover, some variables had only few measurements in the baseline phase, making it difficult to draw reliable conclusions. Narrowing down the variables of interest and measuring them more often could solve this problem. The measurement instruments we used are another concern. Specifically, the newly developed questionnaire of body awareness should be thoroughly validated to establish its usefulness beyond the present study. For emotion regulation, there exists a short form of the DERS (Kaufman et al., 2016), which we were unaware of during the preparation of the study. Likewise, the brief measure of the Big Five is not as reliable as longer scales (Rammstedt & John, 2007) and should be substituted by more reliable forms. In addition, future studies could complement subjective measures with more objective ones, such as heart-rate variability or breath counting (Levinson et al., 2014). We expounded results on the subjective experiences and adverse events related to the four meditation techniques in another manuscript (Matko & Sedlmeier, unpublished). Furthermore, we conducted qualitative interviews with half of our sample and will report our findings in future publications.

In addition, the size of the four subgroups might have been too small to detect between-group differences due to insufficient statistical power. Furthermore, the treatment might have been too short and the setting too similar to establish meaningful differences. Differences between active treatments tend to be small (Wampold et al., 1997), and studies with larger sample sizes might also expose smaller but meaningful differential effects and substantiate reliable predictors of treatment outcome. Future studies could examine whether our findings generalize to other meditation techniques (Matko & Sedlmeier, 2019). Furthermore, they could include an active non-meditation control group (e.g., Hunt et al., 2018) to investigate whether the effects are specific to meditation. All of these efforts could contribute to developing a future theory of meditation and understanding how and for whom it works.