2.1. Participants
In total, 635 participants completed the online survey. Only persons who met the following criteria were asked to complete the survey: all participants (1) must be at least 18 years and no more than 75 years old, and (2) must be fluent in English. The upper limit of age was set, as lucid dream incidences and cognitive capacities have been shown to decline over age [
39]. The convenience sample splits up into a German student population from Osnabrück University (Uos,
n = 72) and students from the University of Texas at Austin (UT Austin,
n = 272), as well as a general mixed international sample of 291 respondents. The study was approved by the Institutional Review Board of the University of Texas at Austin (STUDY00003582). Notably, all participants who claimed to have experienced at least one lucid dream had to verify their understanding of the lucid dream experience. Participants who did not pass the verification were not eligible and were thus excluded. All respondents were grouped depending on their meditation frequency [
40,
41]. Participants were either classified as non-frequent meditators (i.e., meditating less than once per week), as weekly frequent meditators (WFMs; meditating once or more per week), or as daily frequent meditators (DFMs; meditating at least twice per day or multiple times daily).
Within the Osnabrück University sample, 66 participants were eligible (49 females, 16 males, and 1 non-binary; age = 22.94 ± 5.6 (M ± SD)). Based on meditation frequency, 2 students were classified as DFMs, 11 meditated weekly, and 53 were non-frequent meditators. Within the student population from UT Austin, 241 were classified as eligible (156 female, 81 male, and 4 non-binary; age = 19.43 ± 1.89 (M ± SD)). Based on meditation frequency, 1 student was classified as DFM, 29 meditated weekly, and 211 were non-frequent meditators. Within the general mixed population, out of 291 initial respondents, 270 were eligible, more male respondents completed the survey (112 females, 149 males, 4 non-binary, and 5 self-described, e.g., “Genderfluid”), and their ages ranged from 18 to 75, with an average of 37.74 ± 16.16 (M ± SD). Within the general mixed sample, 35 participants were DFMs, 117 were WFMs, and 118 were classified as non-frequent meditators.
2.3. Measures
In order to achieve a comprehensive assessment while maintaining time efficiency, shortened versions of many instruments were implemented. For internal consistency of the scales, McDonald’s omega total was preferred over Cronbach’s alpha. Since it can be assumed that not all items contribute equally to their score, McDonald’s omega is a more accurate reliability estimate, especially for multidimensional or ordinal scales [
42,
43,
44]. Omega can be described as the proportion of variance in observed scores that can be attributed to a single underlying factor or to the common variance among the items on a scale [
45,
46]. As with Cronbach’s alpha, larger values indicate a higher reliability [
47].
Dream Recall and Lucid Dreaming Experience. Lucid and ordinary dream experiences were recorded with an adapted dream survey. The original questionnaire developed by Baird et al. [
48] was modified to fit the specific aims of this study. All participants reported dream recall frequency and lucid dreaming frequency on a 16-point Likert scale, extending the established scales by Schredl and Erlacher [
49]: 0 =
never; 1 =
less than 1 (lucid) dream per year; 2 =
1 (lucid) dream per year; 3 =
2 (lucid) dreams per year; 4 =
3–5 (lucid) dreams per year; 5 =
6–8 (lucid) dreams per year; 6 =
9–11 (lucid) dreams per year; 7 =
1 (lucid) dream per month; 8 =
2 (lucid) dreams per month; 9 =
3 (lucid) dreams per month; 10 =
1 (lucid) dream per week; 11 =
2 (lucid) dreams per week; 12 =
3–4 (lucid) dreams per week; 13 =
5–6 (lucid) dreams per week; 14 =
1 (lucid) dream per night; 15 =
more than 1 (lucid) dream per night. Based on the methodology of Stumbrys, Erlacher, and Malinowski [
33], class means transformed the ordinal scores into metric frequencies either as units per month (0 → 0, 1 → 0.0714, 2 → 0.0833, 3 → 0.1667, 4 → 0.3333, 5 → 0.5833, 6 → 0.8333, 7 → 1, 8 → 2, 9 → 3, 10 → 4, 11 → 8, 12 → 13.5, 13 → 23.5, 14 → 30, 15 → 33) or units per week (0 → 0, 1 → 0.0185, 2 → 0.0192, 3 → 0.0385, 4 → 0.0769, 5 → 0.1346, 6 → 0.1923, 7 → 0.25, 8 → 0.50, 9 → 0.75, 10 → 1, 11 → 2, 12 → 3.5, 13 → 5.5, 14 → 7, 15 → 9). The same class means-recoded 16-point scale was given for lucid dream induction frequency per month. Participants received a written definition along with the scales: “Lucid dreaming is a special sort of dream in which you know that you are dreaming while still in the dream. Typically, you tell yourself “I’m dreaming!” or “This is a dream!”. In some cases, you may also control the content of the dream and alter the dream events as well as your actions voluntarily”.
Respondents who had previously experienced lucid dreams were asked detailed questions regarding their lucid dream experiences, their ability to control lucid dreams, and their training in lucid dream induction techniques. In addition to the monthly lucid dreaming frequency, the number of lucid dreams in the previous six-month period was assessed, which is a summative measure (i.e., an overall measurement taken after a period of time has passed) of lucid dreaming frequency as opposed to a formative approach (i.e., a measurement at shorter time intervals for each week or month). All items were presented either as an open text field or as a Likert-type format. The following single items were used: success of the lucid dream induction: “If you decide to have a lucid dream on a given night, how likely will you succeed?” (0 =
very unlikely; 4 =
very likely); wake-initiated lucid dream occurrences (0 =
never; 4 =
always); and how often one experiences a detached observer stance in the lucid dream (0 =
never; 4 =
always). To verify participants’ understanding of the lucid dream state, they were required to provide a brief report of one of their lucid dreams, detailing how they realized that they were dreaming. As all scales were adapted or created for this study, traditional reliability measures were not applicable. Nonetheless, a strong correlation was found between the frequency of lucid dreams per month and the number of lucid dreams in the previous six-month period (r
sp = 0.93,
p < 0.0001). Participants who reported experiencing lucid dreams at least once per month were categorized as monthly frequent lucid dreamers (MFLDs), while those who reported experiencing lucid dreams at least once or more per week were classified as weekly frequent lucid dreamers (WFLDs), extending the standard classification convention [
9,
29,
50,
51].
Meditation Experience and Frequency. A revised version of the Meditation Experience Questionnaire [
28] was utilized to assess the quantitative experience of meditation practices. To cover various meditation frameworks, three options of Buddhism (Theravadan, Tibetan, or Mixed) were extended to 18 different meditation frameworks: 9 religious/spiritual-oriented traditions (Theravadan, Tibetan, and Zen Buddhism; Daoism; Yoga; Sufism; Judaism; Christianity; and Shamanism) and 9 secular-oriented frameworks were included (app-guided, online-based, Vipassana, self-guided, Yoga, Thai Chi/Qigong, MBSR-based, non-dual meditation, and Transcendental Meditation), plus the option to specify an individual framework and tradition. Two items assessed previous meditation experience (
yes/
no), and meditation frequency. Meditation frequency was measured with a 16-point scale (0 =
never, 15 =
more than 1 meditation per day). Class means transformed the ordinal scale into metric units per week (0 → 0, 1 → 0.0185, 2 → 0.0192, 3 → 0.0385, 4 → 0.0769, 5 → 0.1346, 6 → 0.1923, 7 → 0.25, 8 → 0.50, 9 → 0.75, 10 → 1, 11 → 2, 12 → 3.5, 13 → 5.5, 14 → 7, 15 → 9). If participants meditated at least once per week, they reported how many different techniques they used in their meditation practice on a regular basis. In addition to questions about the quantitative meditation routine, they reported on the styles practiced while meditating, as well as meditation training and retreat experience. For each regularly practiced meditation framework, respondents provided an estimate of the length of an average meditation session in minutes, the frequency per day, the number of days per week, and the number of years of practice within the meditation framework. Based on this account, the total number of hours spent in meditation for each framework per week was calculated:
. Moreover, the participants indicated the percentage of the total time they dedicated to each meditation framework in which a specific quality was facilitated: “Please indicate what percentage of your average meditation time you spend on a specific meditation technique”. Respondents were presented with six options: open monitoring (OM) meditation, focused attention (FA) meditation, loving-kindness/emotionally toned (LK) meditation, meditation to recognize the nature of the mind, non-dual meditation, and one option for an individually specified technique.
Meta-Awareness. The Meta-Awareness subscale of the Multidimensional Awareness Scale (MAS) captured the cognitive ability to recognize one’s current mental state based on self-assessment [
52]. The item “I am aware of my thoughts and feelings as I experience them” reveals the direct aim of measuring the trait aspects of the cognitive process, as the instruction asked participants to indicate the extent to which the given statements represent the typical experience of their thoughts or feelings. The MAS-MA subscale consists of 7 items selected from the original 25-item MAS scale. The items were rated on a 7-point Likert scale (1 =
strongly disagree, 7 =
strongly agree). Reliability was found to be good,
ωt = 0.89.
Trait Mindfulness. The measurement of trait mindfulness was performed using multiple instruments: the Toronto-Mindfulness Scale (TMS) [
53,
54], a short version of the Five Facet Mindfulness Questionnaire (FFMQ) [
55,
56], and the Relaxation, Meditation, and Mindfulness Experience Questionnaire (RMMtm) [
36,
57].
The TMS is a 13-item mindfulness inventory constructed as a 5-point Likert-scale (0 =
not at all; 4 =
very much). The original scale measured state mindfulness and the survey incorporated the trait version of the TMS, asking for experiences during the previous 7 days [
53]. The scale was derived from a two-component definition of mindfulness: Curiosity and Decentering [
1]. Curiosity includes 6 items and refers to the openness to explore one’s internal states as they occur. Decentering, measured by 7 items, refers to the ability to maintain a stance of detachment from one’s thoughts and emotions, with the capacity not to be carried away by thoughts and emotions [
54]. The scales were found to be reliable, with
ωt = 0.91 for Curiosity and
ωt = 0.85 for Decentering.
Drawing from various definitions of mindfulness, a psychometrically validated trait mindfulness measure, the FFMQ, was developed. It combines five mindfulness instruments, and factor analysis on a large sample revealed five factors [
58]. Its condensed form has 20 items rated on a 5-point Likert scale (1 =
never or rarely true, 5 =
very often or always true) [
56]. The brief version measures five distinct but related components of mindfulness, with 4 items for each scale: (1) Observing: the ability to notice and attend to internal and external experiences,
ωt = 0.76; (2) Describing: the capacity to articulate one’s experiences in words,
ωt = 0.77; (3) Acting with awareness: as opposed to “being on autopilot”, the degree of presence and awareness while engaged in activities,
ωt = 0.88; (4) Nonjudgment of inner experience: the ability to refrain from judging inner experiences as neither good nor bad,
ωt = 0.88; (5) Nonreactivity to inner experience: the attitude of allowing thoughts and feelings to arise and pass without being caught up in or swept away by them, allowing these experiences to exist without interference,
ωt = 0.77. The psychometric properties of the short version of the FFMQ by Tran et al. [
56] demonstrated moderate to strong reliability.
The Relaxation, Meditation, and Mindfulness Tracker (RMMtm) by Smith [
59] comprises a self-reported broad-spectrum inventory for assessing dispositional mindfulness. The 7-point Likert scale with 32 items captures trait mindfulness (1 =
never, 2 =
not this month, but once or twice this year; 3 =
about once this month; 4
= about once a week; 5 =
about 2 or 3 times a week; 6
= about every day; 7 =
several times daily). The RMMtm includes various experiences associated with the practice of mindfulness meditation techniques that manifest as characteristics over time. The instrument was derived from third-wave mindfulness theory to capture all essential phenomenological states of mindfulness, represented on a continuum with varying levels of mindfulness [
60]: (1) Mindful Relaxation, which assesses initial mindful relaxation experiences; (2) Mindful Quiet Focus, which captures the stillness and observational nature of meditation; (3) Mindful Engagement, which measures the ability to remain present and compassionate during activities; and (4) Mindful Transcendence, the deepest level observed in long-term practitioners in spiritual or non-secular contexts [
61]. Due to expected variations in levels of mindfulness among long-term and frequent meditators compared to other groups, separate component analysis for each population sample is recommended [
36,
57]. A principal component analysis identified 2 distinct factors. Items 1–24 loaded on the first factor; this dimension is interpreted as Mindful Relaxation and Focus,
ωt = 0.97. The first dimension includes aspects of physical relaxation but also cognitive and emotional aspects of mindfulness: “I was living in the present moment, not past or future concerns” or “I felt selfless/caring/compassion”. Items 25–32 loaded on the second factor, compromising Mindful Transcendence,
ωt = 0.93. Self-transcendence, in general, is defined as the capacity to expand self-boundaries [
62]. The subscale refers to transpersonal self-transcendence: “I had a sense of what is timeless, boundless, infinite”. Due to the sample-specific component analysis, reliability measures were not comparable to previous research. However, reliability analysis indicated satisfying internal consistencies.
2.4. Statistical Analysis
Statistical analyses were performed using R, version 4.3.1, and SPSS, version 29. For the primary data analysis, only the general mixed sample was utilized, following manual validation of each participant’s lucid dream report. The student populations were not included in the data analysis for the research questions addressed in this study, since there was an insufficient number of participants per group. Data management was based on the functions of the tidyverse package in R [
63]. A total of 270 participants out of 291 were considered eligible for data analysis. The more liberal Shapiro–Wilk tests revealed significant deviations from normal distribution for most of the measures, e.g., lucid dreaming frequency (Shapiro–Wilk’s W = 0.590,
p < 10
−14), dream recall frequency (Shapiro–Wilk’s W = 0.855,
p < 10
−15), lucid dream induction frequency (Shapiro–Wilk’s W = 0.540,
p < 10
−22), and meditation frequency (Shapiro–Wilk’s W = 0.813,
p < 10
−16). Therefore, non-parametric independent two-sample permutation tests were utilized for assessing overall group differences. For each group, a set of multiple tests with adjusted
p-values based on Benjamini and Hochberg and the False Discovery Rate (FDR) correction were reported [
64].
Monte Carlo permutation tests with R = 10,000 permutations were implemented for group comparisons [
65]. For most group comparisons, the Monte Carlo permutation test of the mean was calculated. All variables besides the ordinal scaled WILD frequency and the ordinal variable related to the detached observer stance were continuous variables. When comparing the total hours of meditation per week and the largest number of years for a meditation practice, a Monte Carlo permutation test of the median was used. In addition to that, Spearman’s correlation was preferred over the Pearson correlation due to the influence of the largest values in the weekly hours of meditation for each meditation practice style.
Prior to this, an exploratory principal component analysis found the sample-specific RMMtm scales [
36]. Bartlett’s test of sphericity was significant (
X2 = 8357.33, df = 496,
p < 0.001), indicating that the variables were sufficiently intercorrelated to proceed with principal component analysis (PCA). A PCA with Kaiser normalization and an Oblimin rotation method yielded a two-factor solution in the general population with tools from the psych package [
66]. The Kaiser–Meyer–Olkin Measure was 0.953, which indicates good sampling adequacy. Factor extraction was based on the scree plot, indicating two factors, while parallel analysis yielded two factors and Kaiser–Gutman criteria indicated four factors. The two-factor solution accounted for 62% of the cumulative variance, compared to 67% for the three-factor and 70% for the four-factor solution. The tools from the MBESS package calculated omega total (
ωt) estimates instead of the psych tools for a more conservative reliability measure [
43,
47].
Multiple regression analysis for the monthly lucid dreaming frequency, predicted by the RMMtm Mindful Transcendence subscale, the TMS Decentering and Curiosity subscale, the MAS-MA subscale, weekly meditation frequency, and age, was implemented with the Boot and LessR package in R (i.e.,
Table S1, Figures S1 and S2) [
67,
68]. Due to violations of heteroscedasticity and normal distribution, examined by means of the visual plotting of the predicted values against the standardized residuals, the coefficients were tested based on 10,000 bootstrapped bias-corrected confidence intervals. Assumptions were investigated via the LessR package: no VIF values exceeded 5, and none of Cook’s distance indexes was larger than 1, with the highest VIF being 3.634 and the largest Cook’s distance index being 0.12.
The exploratory analysis of the relationship between lucid dreaming, meditation, and meta-awareness was performed with model 4 of the PROCESS macro for R [
69]. In this model, meta-awareness was used as a mediator, meditation frequency as the independent variable, and the number of lucid dreams in the previous six months was used as the outcome variable. The model controlled for covariates including lucid dream induction frequency, dream recall frequency, and age. It must be stated that for this analysis, 25 participants who did not have prior experience with lucid dreams were excluded, due to missing data. Bias-corrected accelerated (BCa) confidence intervals for the coefficients were bootstrapped with R = 10,000 replicates, as was the confidence interval for the indirect effect. Standard errors were computed using heteroscedasticity-consistent estimates due to the heteroscedasticity of the residuals and the nonnormality of the dependent variable (Shapiro–Wilk’s W = 0.381,
p < 10
−22). The model parameters were standardized, and the random seed was fixed to 9999 to ensure the reproducibility of the results.