Mindfulness Predicts Mind Wandering and Differentially Affects Its Controllability in Everyday Life

ABSTRACT

It is currently unresolved whether control, or its lack, over mind wandering contributes to its associations with creativity. Mind wandering and mindfulness are seen as opposing constructs concerning the ability to remain undistracted and, at the same time, appear to have a complex relationship with one another and with creativity. This study investigated the influence of control on novelty and usefulness of thought content during mind wandering in everyday life, controlling for mindfulness, creativity, and momentary mood. One thousand twenty-five thought probes of 41 healthy young adults were collected over the course of one week. The mindfulness facet acting with awareness and higher originality predicted less mind wandering. Acting with awareness also predicted higher, but nonreactivity to inner experience and creative production lower, momentary control. Momentary control was lower among women than men and positively associated with usefulness, but not novelty, of thoughts. Creativity did not moderate these associations. Habitual spontaneous mind wandering predicted lower usefulness and novelty of thoughts. These results highlight the need to distinguish controlled and uncontrolled mind wandering and the facets of mindfulness in this field of inquiry. Future research should focus on the contents and processes of mind wandering to further disentangle its adaptive and maladaptive consequences.

Introduction

Mind wandering is a universal human experience. Smallwood and Schooler (2006) define it as a shift of attention away from external, task-related information to internal, task-unrelated information. However, this definition is so broad that various conceptualizations have been offered for it in terms of underlying dynamics (spontaneity), content (task-unrelated thought, autobiographical memory), and its mode of presentation (daydreaming). This suggests a so-called definitional haze surrounding mind wandering (Christoff et al., 2018; but see; Seli et al., 2018), which poses a hindrance to empirical research progress. Nevertheless, studies have estimated human minds to wander from 15–25% (Murray, Irving, & Krasich, 2022; Smith et al., 2022) to 30–50% of the day (Kane et al., 2017; Killingsworth & Gilbert, 2010; Seli et al., 2018). For its distractive qualities, mind wandering is associated with maladaptive outcomes in many different situations, such as driving, reading, and during lectures (Mooneyham & Schooler, 2013). Yet, it could also make boring moments useful (Feng, D’Mello, & Graesser, 2013) and could have utility for autobiographical planning (Baird, Smallwood, & Schooler, 2011) and social problem solving (Ruby, Smallwood, Sackur, & Singer, 2013).

Most relevant to the goals of the current study, mind wandering seems to boost creativity (Agnoli, Vanucci, Pelagatti, & Corazza, 2018), and there also are similarities in the neuronal substrate of mind wandering and creativity (both involving the medial temporal lobe and the default network; for a review, see Fox & Beaty, 2019). However, extant empirical evidence on relationships between mind wandering and creativity is inconsistent (Agnoli, Vanucci, Pelagatti, & Corazza, 2018; Hao, Wu, Runco, & Pina, 2015).

This inconsistency is likely attributable to the varying definitions and operationalizations of these two constructs used in research (see also Murray, Irving, & Krasich, 2022). Both mind wandering and creativity are complex constructs, the measurement of which is challenging. These issues are discussed in the following three sections. Additionally, there also appears to be a complex relationship between mind wandering, creativity, and mindfulness, which may further complicate the investigation of the associations between mind wandering and creativity. This issue is discussed in more detail in a fourth section.

Operationalization of mind wandering

Early studies using self-report data (Singer & McCraven, 1961) already suggested both intraindividual and individual differences in the frequency and characteristics of daydreaming (considered one mode of presentation of mind wandering in the present study). Daydreaming occurred when participants were alone, especially before sleeping, concerned practical problem solving, but sometimes also bizarre and unrealistic scenarios, and decreased with age (a finding recently replicated by McVay, Meier, Touron, & Kane, 2013). Especially interesting for the current study, Singer and McCraven (1961) already observed higher creativity in story telling associated with higher reported daydreaming frequency. Self-report methods are economic and easy to use, but rely on participants’ accurate perception and their willingness, as well as on their ability to report on their experiences in a reliable and valid way. Accordingly, various forms of response bias (and among these, especially recall bias, due to the typically retrospective nature of assessment) may compromise the validity of results (see Weinstein, 2018).

In lab settings, the frequency of mind wandering can be measured directly with the probe-caught method (stopping participants during a specific task and then asking them where their attention was precisely directed at; Feng, D’Mello, & Graesser, 2013), the self-caught method (participants give a sign or push a button every time they notice their attention shifts away from the ongoing task; e.g., Mrazek, Smallwood, & Schooler, 2012), or indirectly with errors and reaction times in the SART (Sustained Attention to Response; a Go/No-go task; Mrazek, Smallwood, & Schooler, 2012). All these methods may limit recall bias (for a review and empirical data comparing different methods, see Kane, Smeekens, Meier, Welhaf, & Phillips, 2021), but pose problems concerning external validity, as participants may act differently in everyday life.

One method to measure mind wandering in everyday life is the experience sampling method (ESM). ESM is a diary method that measures situational phenomena in natural environments, prioritizes concrete and immediate experiences over abstract and recalled experiences, and provides sampling of experiences in daily life (Beal, 2015). Consequently, ESM is characterized by high external validity and general reductions of recall, memory, and method biases as well (Beal, 2015; Iliev, Medin, & Bang, 2022). Even though the frequency of mind wandering in lab settings correlates with the frequency of mind wandering in everyday life and is associated with performance problems in both contexts (McVay, Kane, & Kwapil, 2009), most people these days have their smartphone at hand all day, which makes the smartphone the perfect ESM device.

ESM thus has been frequently used in studies on mind wandering. One such investigation asked participants at five random times during the day for 1 week about their mood and whether they had mind wandered in the moment before the signal (Franklin et al., 2013). If mind wandering occurred, participants rated their thought content in terms of perceived novelty, usefulness, and interest. Both interesting and useful mind wandering content were related to positive mood. Yet, participants generally reported lower mood during mind wandering; a finding that was also reported in an ESM study by Killingsworth and Gilbert (2010), which led them to the conclusion that mind wandering causes low mood. Other ESM studies provided evidence that mind wandering only preceded low mood (Poerio, Totterdell, & Miles, 2013). Conclusions about causality should therefore be made with caution.

Controlled vs. uncontrolled mind wandering

Mind wandering can be differentiated based on the amount of control (“controlled vs. uncontrolled” are used interchangeably here with the alternative terms “intentional vs. unintentional” and “deliberate vs. spontaneous”) one has over its initiation and process (Carriere, Seli, & Smilek, 2013). The content of controlled mind wandering is more future-oriented and concrete (Seli, Ralph, Konishi, Smilek, & Schacter, 2017), whilst past-oriented mind wandering is associated with lower mood than future-oriented mind wandering (Poerio, Totterdell, & Miles, 2013). Perseverative or ruminative forms of mind wandering are among its uncontrolled forms and are associated with adverse conditions, such as depression and anxiety (Ottaviani & Couyoumdjian, 2013).

Mind wandering and creativity

In the multidisciplinary literature on creativity, many definitions are encountered, chiefly having two aspects in common: Creativity requires originality/novelty and effectiveness/utility (Runco & Jaeger, 2012). While some operationalizations of creativity attempt to measure individual creative achievement in terms of usefulness to society (“big c;” valued, admired, and used by others), others target everyday creativity (“little c;” used to solve practical problems in everyday life; Kaufman & Beghetto, 2009). In the context of intelligence research, creativity is defined as the ability to think divergently, i.e., to approach problems in an open, unsystematic, and playful manner, thus eliminating constraints in thinking (Guilford, 1968).

There are interesting parallels between creative thinking and mind wandering: Both can be described by a dual process model consisting of production and evaluation, and both apparently have similar neural substrates (Fox & Beaty, 2019). In addition, both creativity and mind wandering can be evaluated in terms of perceived novelty and usefulness. Yet, while divergent thinking can be facilitated by associative processes via episodic memory reactivation (McCormick, Rosenthal, Miller, & Maguire, 2018), mind wandering during a divergent thinking task may lead to a decrease of performance (Hao, Wu, Runco, & Pina, 2015; Murray, Liang, Brosowsky, & Seli, 2021). Mind wandering is a resource-consuming process and may thus interfere with creative idea production. Thus, it appears crucial to be able to control mind wandering to foster creative processes.

Fox and Beaty (2019) have proposed that exceptionally creative and inventive individuals do not differ from the average population (or psychiatric groups) in the frequency of mind wandering, but rather in the frequency of novel and useful thoughts. According to this model, higher creativity comes with more control over the content of thoughts during mind wandering and the ability to channel thoughts into a creative (i.e., novel and useful) direction. At the other end of this continuum are individuals with psychiatric diagnoses (such as depression or post-traumatic stress disorder), whose mind wandering is characterized by ruminative patterns. The content is not new, but rather repetitive and not useful. This kind of mind wandering is perceived as unpleasant and involuntary. Consistent with this model, spontaneous, but not deliberate, mind wandering has been associated with attention-deficit hyperactivity disorder and obsessive-compulsive disorder (Seli, Smallwood, Cheyne, & Smilek, 2015, 2017).

Mindfulness

Mindfulness (“paying attention in a particular way, on purpose, in the present moment, and nonjudgmentally;” Kabat-Zinn, 1994, p. 4) is a construct highly relevant for clinical contexts (e.g., Goldberg, Riordan, Sun, & Davidson, 2021), and in addition is associated with a host of other psychological constructs, such as prosocial behavior or motivation (Donald et al., 2018, 2021). With respect to the ability to remain undistracted, mind wandering and mindfulness can be considered opposing constructs (Smallwood & Schooler, 2006). Consequently, mindfulness meditation diminishes mind wandering in the SART paradigm (Mrazek, Smallwood, & Schooler, 2012). However, mindfulness is also linked to higher creativity (with a meta-analytic average correlation of r = .22; Lebuda, Zabelina, & Karwowski, 2016) and may be further associated with creative performance and divergent thinking via an interaction effect with controlled mind wandering (Agnoli, Vanucci, Pelagatti, & Corazza, 2018).

The Five Facet Mindfulness Questionnaire (FFMQ; Baer, Smith, Hopkins, Krietemeyer, & Toney, 2006) is one of the most widely used measures of mindfulness and operationalizes mindfulness as a five-facetted construct (observing, describing, acting with awareness, nonjudging of inner experience, nonreactivity to inner experience). Interestingly, these facets appear to be differentially associated with habitual spontaneous and deliberate mind wandering (Seli, Carriere, & Smilek, 2014). Spontaneous and deliberate mind wandering are both positively associated with observing (i.e., the propensity to observe and/or attend to one’s sensations, perceptions, thoughts, and feelings). Yet, spontaneous mind wandering is negatively associated with all other facets as well (strongest with acting with awareness, i.e., the tendency to act carelessly or in an “auto pilot” style, followed by nonjudging of inner experience and least with describing and nonreactivity to inner experience, i.e., the ability to “step back” and not reacting immediately or automatically to sensations and experiences), whereas deliberate mind wandering with only acting with awareness and nonjudging of inner experience (Seli, Carriere, & Smilek, 2014).

These differential associations suggest that conflicting or inconsistent associations of mind wandering with mindfulness in extant research may either originate from conflating spontaneous and deliberate mind wandering, be due to not accounting for the various facets of mindfulness, or both. Mind wandering and mindfulness apparently have a complex relationship and, beyond that, with creativity as well (Agnoli, Vanucci, Pelagatti, & Corazza, 2018, highlighted the associations of describing, acting with awareness, and nonreactivity to inner experience with creativity, but not of the other mindfulness facets). Therefore, it may be worthwhile to control for mindfulness and its facets in investigations into the associations of mind wandering with creativity as well.

Present study

This study investigated associations between mind wandering and the production of creative (i.e., novel and/or useful) thoughts in everyday life in a sample of healthy participants with the aim to overcome shortcomings of previous research and to arrive at potentially clearer conclusions. To achieve these goals, ESM was used for the assessment of mind wandering and thought content and the perceived control over mind wandering (both habitual and momentary) as predictors of creative production during mind wandering, whilst controlling for the facets of mindfulness, different aspects of creativity (both self-reported and assessed with objective methods), and momentary mood. Thereby, the individual contributions of these factors were sought to be disentangled, given previous mixed and unclear findings. Creativity was assessed with different methods, which tapped into different aspects of the construct, given its variable conceptualization and operationalization in extant research. Mind wandering was defined as “thoughts arising regardless of the current activity or perception.” The assumptions that higher mindfulness is associated with less mind wandering (Research Hypothesis [RH] 1a) and higher momentary control over mind wandering (RH 1b) were tested first. Second, it was investigated whether more control predicted more creative thoughts, operationalized via higher perceived usefulness (RH 2a) and higher perceived novelty (RH 2b) of the thought content. Third, it was investigated whether higher creativity moderated (i.e., strengthened) the relationships between perceived control and usefulness (RH 3a) and novelty (RH 3b) of thought content. This tested the prediction of Fox and Beaty’s (2019) model that higher creativity comes with more control over the content of thoughts during mind wandering and the ability to channel thoughts into a creative (i.e., novel and useful) direction. Figure 1 is a graphical representation of the conceptual model underlying the nexus of the present study variables. In sum, this study tested important predictions from previous research, concerning the role of mindfulness, and from the recent model of Fox and Beaty (2019), concerning the role of creativity and control on the production of creative thoughts during mind wandering. Participant sex was also controlled for in analysis, as potential sex differences are underinvestigated in this field of research.

Figure 1. Schematic representation of the conceptual model and the research hypotheses (RH).

See the main text for details.

Materials and methods

Participants

The sample consisted of 41 healthy young adults aged 20–29 years (M = 24.3, SD = 2.2), including 29 women and 12 men. Of the sample, 30 participants (73%) were currently students, and a total of 21 individuals (51%) had completed some tertiary education. Participants were of German, Austrian, or other Central European nationality (n = 30, 8, and 3; equivalent to 73%, 20%, and 7%, respectively). Inclusion criteria were a minimum age of 18 years and possessing a smartphone capable to receive on-screen notifications of a mail application. Recruitment took place via notices in buildings of the University of [masked for review] and the University of Applied Arts [masked for review], and via social media networks. Participation was voluntary and without incentives. Participant data were anonymized for analysis.

Material

Habitual mind wandering

Everyday mind wandering was assessed with the Mind Wandering Deliberate (MWD) scale and the Mind Wandering Spontaneous (MWS) scale, which each consist of four items with a 7-point Likert response scale (Carriere, Seli, & Smilek, 2013). Previous studies have obtained good reliability and discriminant validity for the English and Italian versions of these scales (Agnoli, Vanucci, Pelagatti, & Corazza, 2018; Seli, Carriere, & Smilek, 2014). For the present study, the scales were translated into German using the parallel-blind technique (Behling & Law, 2000). Internal consistency (Cronbach α) of the scales in the present sample were .83 (MWD) and .70 (MWS).

Creativity

Creativity was assessed with three measures (CAQ, AUT, and ADT), in order to capture different aspects of this construct. Measuring only single aspects may provide an incomplete picture; at the same time there is evidence that creative achievements are also domain-specific (for an overview and discussion, see Said-Metwaly, Van den Noortgate, & Kyndt, 2017). The Creative Achievement Questionnaire (CAQ; Carson, Peterson, & Higgins, 2005) is a checklist of creative achievement with 96 items and focuses on visible and outstanding creative achievements (“big c”) in 10 domains (visual arts, music, dance, architectural design, creative writing, humor, invention, scientific discovery, theater and film, culinary arts). Participants state which out of seven rank-ordered indicators of creative achievement per domain apply to them. Some items also query the number of times this applied to them in past (e.g., the number of times participants’ compositions have been critiqued in a national publication). Indicators are scored from 1 to 7 within each domain (if none applies, a score of 0 is awarded; for indicators, which assess the number of achievements, scores are multiplied by this number) and summed across all domains. The CAQ has been found to have good validity, as evidenced through its correlations with established measures of creativity, such as divergent thinking (r = .47; Silvia, Wigert, Reiter-Palmon, & Kaufman, 2012). Sample reliability (Cronbach α) of the CAQ was .87.

The Alternative Uses Task (AUT; Guilford, 1967) is a widely used test to assess divergent thinking. Respondents have 2 minutes to think of an alternative use for a familiar object. For the present research, the object chosen was a tissue. Responses were scored concerning two criteria: Fluency (i.e., number of responses) and Originality (i.e., rarity of responses). Fluency is scored as a participant’s total number of responses. Scoring Originality involved comparing each individual response to its prevalence in the entire sample (following Blanco-Herrera, Gentile, & Rokkum, 2019; for an overview of widely used alternative scoring methods, see also Reiter-Palmon, Forthmann, & Barbot, 2019). Responses given by only 2–5% of participants scored one point.

The Alien Drawing Test (ADT; Kozbelt & Durmysheva, 2007) measured creative production. Participants were given 7 minutes to imagine and draw a creature that evolved on a planet very different from Earth. The result was scored on three criteria: The quantity of eyes, the quantity of limbs, and the presence of bilateral symmetry. Creatures with two eyes, four limbs, and bilateral symmetry scored zero points in each category. Creatures with more or fewer than two eyes, more or fewer than four limbs, and superficially violated bilateral symmetry (e.g., one extra limb on one side) scored one point per category. Creatures with no eye structure, no limb structure, and no symmetry scored two points per category. Because the ADT requires immediate generation of a creative product, it is considered a valid instrument for measuring creative production. In the present study, interrater agreement of two raters (Cohen κ, with 95% confidence intervals) was .80 [.64, .96], .65 [.44, .87], and .46 [.23, .70] for eyes, limbs, and symmetry, respectively (percent agreement was 88%, 80%, and 68%). These numbers were comparable to, and their confidence intervals included, the interrater agreements reported by the authors of this scale themselves (Kozbelt & Durmysheva, 2007, p. 84, .78., and .62, respectively). Final ratings were arrived at by discussion and consensus.

Mindfulness

Mindfulness was assessed with a validated German short form (Tran, Glück, & Nader, 2013) of the Five Facets Mindfulness Questionnaire (FFMQ), comprising 20 items with a 5-point Likert response scale with endpoints 1 (never or very rarely true) and 5 (very often or always true). The FFMQ is composed of five subscales, each capturing different aspects of everyday mindfulness: Observe, Describe, Acting with Awareness (Actaware), Not Judging Inner Experience (Nonjudge) and Being Nonreactive to Inner Experience (Nonreact). In the present sample, Cronbach α was, in above order, .68, .72, .70, .80, and .65.

Experience sampling protocol

The experience sampling protocol was implemented as signal contingent at random intervals, using participants’ smart phones and the online survey tool SoSciSurvey (Leiner, 2019). Data collection took place in March and April 2020 during the first wave of the current COVID-19 pandemic, under conditions of the first nationwide lockdown in Austria and Germany (Müller, 2020). The survey period for participants spanned 1 week (Monday through Sunday). Participants received a thought probe via mail five times per day. For this, all participants had activated the on-screen notifications of their mail account on their smartphones and, whenever possible, had also turned on the sound or vibration mode, in order to notice to the prompt immediately. Participants were instructed to react immediately to this prompt whenever possible, but to ignore the notification, if the time delay in their reaction exceeded 30 minutes. Individual thought probes were sent to participants at least 30 minutes apart. The questions in the ESM referred to the moment just before the request was noticed (i.e., heard, sensed, or seen on the smart phone). With regard to this requirement, all participants were briefed and trained in detail. Each thought probe posed two to five questions. The first question was: “Did your mind wander?” In case of a negative answer, only momentary mood (“What is your current mood?”), ranging from 1 (bad) to 101 (good), was assessed. In case of a positive answer, participants were asked to rate their momentary mood and then their thoughts on how new (“How new were your thoughts?”), how useful (“How useful were your thoughts?”), and how controlled (“How controlled were your thoughts?”) their mind wandering was. These three specific follow-up questions were answered on a 5-point Likert scale, with endpoints 1 (not at all) and 5 (very much).

Procedure

All participants received and signed an informed consent form in an initial, individual face-to-face meeting, in which they were informed about the content of the study, their rights, and data protection details. Subsequently, all participants completed a paper-pencil task sheet in which they provided basic demographic information and completed the MWD, MWS, FFMQ, CAQ, AUT, and ADT. One researcher (AN) stopped completion times for the AUT and ADT. This was followed by a detailed briefing and training session for the ESM, wherein definitions of terms, along with technical and procedural details, were clarified. For each participant, the ESM assessment period started on a Monday, following this initial face-to-face meeting.

Almost immediately with the start of the initial, individual assessments, contact restrictions to contain the COVID-19 pandemic were set into place in Austria and Germany (World Health Organisation WHO, 2020). These included the closure of universities and offices, as well as contact restrictions and exit restrictions. Therefore, fewer than 10 participants completed the initial assessment with the test administrator in a face-to-face meeting (in participants’ place of residence). Instead, a video call was used. The video call procedure was similar to the face-to-face meeting (i.e., testing took place in participants’ place of residence). With respect to the COVID-19 pandemic, after completion of the ESM, all participants were asked via e-mail one final question: “How much have the COVID-19 pandemic containment restrictions affected your life in the past week?” Response options ranged from 1 (not at all) to 10 (very much).

All research in this study was performed in accordance with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards, and with institutional guidelines of the [masked for review]. Study participation did not affect the physical or psychological integrity, the right for privacy, or other personal rights or interests of the participants. Such being the case, according to national laws (Austrian Universities Act, 2002), this study was exempt from formal ethical approval.

Statistical analysis

Four models were fitted on the data. Model 1 (addressing RH 1a) predicted the overall likelihood of mind wandering, whereas Models 2 to 4 the momentary control (RH 1b), usefulness (RH 2a), and novelty (RH 2b) aspects, respectively. Participant sex, habitual mind wandering (deliberate and spontaneous), creativity (all aspects), and momentary mood were utilized as Level-2 and Level-1 predictors, respectively, in all four models. FFMQ subscale scores were utilized as further Level-2 predictors in Models 1 and 2. Momentary control, usefulness, and novelty were utilized as further Level-1 predictors in Models 3 and 4 (novelty in Model 3 and usefulness in Model 4), which also included cross-level interactions of momentary control with all individual aspects of creativity (thus addressing RH 3a and 3b). All Level-1 predictors were group-mean centered for analysis (i.e., within participants), whereas all Level-2 predictors grand-mean centered.

Significance was set to p < .05. Restricted maximum likelihood was used for parameter estimation. Because the Level-2 sampling had fewer than 50 units (i.e., study participants), 95% bias-corrected and accelerated (BCa) bootstrap confidence intervals were generated for the MLMs with continuous outcomes, using 5,000 bootstrapped samples. For the parameters in the MLM with binary outcome, the results of Wald-type significance tests (using the Kenward-Roger approximation and a model-based estimator for the covariance matrix) are reported. Unstandardized regression coefficients are reported for all models in order to allow direct comparisons between the models. Odds ratios (ORs) can be computed from the regression coefficients of Model 1 via exponentiation and are provided for significant parameters to allow for direct comparison of the predictors in Model 1. For the metric predictors, ORs are presented for a change by 1 standard deviation (SD) to account for scale differences in the predictors. Further, all significant parameter estimates in Models 2 to 4 are also presented in standardized form b*, by multiplying the unstandardized parameter estimates with the SD ratios of the involved variables (Snijders & Bosker, 2012), i.e., b * SD(X)/SD(Y), using the SDs of the group-mean centered variables, where required. For the dichotomous predictor participant sex, significant parameter estimates are presented in the metric of Cohen d.

Intercepts and the slope of momentary control (in Models 2 and 3) were estimated as random parameters, allowing them to covary across Level-2 units (i.e., participants), where applicable. The variance of these parameters is reported twice: once before adding the Level-2 predictors to the models, and a second time after adding the Level-2 predictors. Thereby, the reduction in variance via adding Level-2 predictors can be directly assessed. For the MLMs with continuous outcomes, significance of the random parameters was tested with likelihood ratio (LR) tests, refitting the models using maximum likelihood estimation and comparing models with and without randomness in these parameters. For the MLM with binary outcome, the provided Wald-type tests were used. As analytic power of these tests of randomness conceivably was low (because of limited sample size), intercepts and slopes (where applicable) were still modeled as random, regardless of the outcome of these statistical tests.

Intraclass correlation coefficients (ICCs) for models without predictors (null models) are reported in order to quantify the strength of clustering in the data descriptively (using formulae provided in Nakagawa & Schielzeth, 2010, for the MLM with binary outcome) and R² values for the MLMs. For all MLMs, conditional R² values (Nakagawa, Schielzeth, & O’Hara, 2013) are reported, which can be interpreted as the proportion of Level-1 variance attributable to fixed and random effects. All analyses were conducted using IBM SPSS version 28. Analysis data and SPSS syntax for the MLM main analysis is provided in OSF at https://osf.io/mvy3n/.

Power analysis

Power analyses were not performed a priori, but only post hoc to evaluate the analytic power of the present study for the presently observed effects. Monte Carlo power calculations (using the Shiny app of Lafit et al., 2020) with the parameter estimates of the present data confirmed that the sample size was large enough to attain sufficient power levels (i.e., ≥80%) for the effect tests of the Level-2 predictors on the random intercept, such as of originality, acting with awareness, and describe in Model 1. However, the test of creative achievement in Model 2 was slightly underpowered (73% power; 20 further participants would have raised power levels to 86%). The test of deliberate mind wandering in Model 4 had a power of 47%. The tests of the cross-level interactions of creative achievement with control in Models 3 and 4 had 42% and 67% power and would have needed at least 20 participants more in Model 4 to achieve sufficient power (83%). As evidenced by the current results (see below), all these associations were, in comparison, of only rather small size and, hence, likely less relevant overall.

Results

One participant, who responded only once to the thought probes, was excluded from analysis. All other participants had individual response rates of ≥28.6% (on average 73.1%; see below). Means and standard deviations of all Level-2 predictors (i.e., person-specific characteristics) and all Level-1 momentary ratings (across all participants and measurement occasions) are provided in Table 1. The response rate in the ESM was 73.1% (1,024 out of 1,400 thought probes). On average, participants responded to 25.6 (SD = 6.1, Range = 10–35) thought probes over the whole study period, amounting to an average of 3.7 thought probes per day (SD = 0.9, Range = 1.4–5). Mind wandering was reported in 37.5% (384 of 1024) of the thought probes. On average, participants reported 9.6 (SD = 4.0, Range = 2–19) episodes of mind wandering over the whole study period.

Table 1. Sample descriptive statistics.

Measure	M (SD)
Person-specific characteristics (Level 2)
Deliberate mind wandering (MWD)	20.18 (5.27)
Spontaneous mind wandering (MWS)	18.47 (4.34)
Creative achievement (CAQ)	9.13 (8.31)
Fluency (AUT)	7.23 (2.18)
Originality (AUT)	1.23 (0.20)
Creative production (ADT)	2.45 (1.38)
Observe (FFMQ)	15.35 (2.66)
Describe (FFMQ)	14.40 (3.18)
Actaware (FFMQ)	11.63 (2.64)
Nonjudge (FFMQ)	13.75 (3.17)
Nonreact (FFMQ)	12.55 (2.43)
Momentary ratings (Level 1)
Mood	69.12 (19.11)
Usefulness	2.63 (1.23)
Novelty	2.62 (1.31)
Control	2.51 (1.19)

MWD = Mind Wandering Deliberate scale; MWS = Mind Wandering Spontaneous scale; CAG = Creative Achievement Questionnaire; AUT = Alternative Uses Task; ADT = Alien Drawing Test; FFMQ = Five Facet Mindfulness Questionnaire.

The intercorrelations of all Level-2 and Level-1 variables that were used in the statistical analysis (utilizing the individual total numbers of mind-wandering episodes and individual means for all other Level-1 variables) are provided in supplementary material (Table S1). For spontaneous mind wandering, the associative pattern with the facets of the FFMQ that was broadly similar to the one reported in Seli, Carriere, and Smilek (2014), i.e., a positive association with observing and negative associations with the four remaining FFMQ facets. However, the pattern differed somewhat for deliberate mind wandering, which had also positive associations with describe, nonjudging of inner experience, and nonreactivity to inner experience in the present data. CAQ scores were higher among men than women. Mind wandering itself (number of episodes relative to the individual number of thought probes) was uncorrelated with all study variables at Level 2. Average usefulness was strongly positively associated with both average control and average novelty. The COVID-19 pandemic containment restrictions had, on average, substantial effects on participants’ lives during the period of data collection (as suggested by M = 7.5, SD = 2.1, on the utilized 10-point scale). Ratings also correlated positively with habitual deliberate mind wandering and average novelty, which likely reflected being confronted with the novel situations of the COVID-19 pandemic and the lockdowns.

Mind wandering (RH 1a)

There was only weak clustering at Level 2 with regard to the probability of mind wandering (ICC = .04; yet, σ²₀₀ only closely missed statistical significance without the Level-2 predictors, p = .056). Mind wandering was more likely (OR = 0.76, 95% confidence interval = [0.61, 0.94]; Table 2) among participants reporting lower originality and was associated with lower momentary mood (OR = 0.80 [0.72, 0.89]). Concerning its association with mindfulness, mind wandering was more likely among participants reporting lower scores in acting with awareness (OR = 0.69 [0.50, 0.95]) and higher scores in describe (OR = 1.33 [1.00, 1.72]; however, this latter association only barely reached statistical significance).

Table 2. Results of the multilevel analyses on the frequency of mind wandering and perceived control, usefulness, and novelty during mind wandering.

Effect	Mind wandering	Control		Usefulness	Novelty
Fixed effects
Intercept	−0.51* [−0.91, −0.10]	2.93 [2.67, 3.19]	2.48 [2.32, 2.58]		2.35 [2.14, 2.51]
Level-1 momentary ratings
Mood	−0.02*** [−0.03, −0.01]	0.01 [−0.0001, 0.01]	0.01 [0.002, 0.01]		0.005 [−0.004, 0.01]
Control	–	–	0.40 [0.30, 0.49]		−0.04 [−0.19, 0.15]
Usefulness	–	–	–		0.17 [0.04, 0.29]
Novelty	–	–	0.13 [0.04, 0.21]		–
Level-2 predictors
Female sex	−0.05 [−0.55, 0.46]	−0.52 [−0.85, −0.20]	0.23 [−0.02, 0.53]		0.32 [−0.02, 0.69]
Habitual mind wandering
Deliberate (MWD)	−0.02 [−0.08, 0.04]	0.01 [−0.03, 0.05]	−0.01 [−0.03, 0.02]		0.03 [−0.002, 0.06]
Spontaneous (MWS)	−0.04 [−0.10, 0.02]	−0.03 [−0.06, 0.01]	−0.04 [−0.06, −0.02]		−0.05 [−0.09, −0.01]
Creativity
Creative achievement (CAQ)	−0.01 [−0.04, 0.02]	0.02 [−0.006, 0.04]	0.03 [0.02, 0.04]		0.02 [−0.002, 0.03]
Originality (AUT)	−1.39* [−2.49, −0.30]	0.18 [−0.46, 0.82]	0.42 [−0.06, 0.81]		−1.24 [−1.85, −0.66]
Fluency (AUT)	0.08 [−0.03, 0.19]	0.002 [−0.07, 0.07]	0.005 [−0.06, 0.08]		−0.01 [−0.08, 0.08]
Creative production (ADT)	−0.05 [−0.20, 0.11]	−0.12 [−0.23, −0.02]	−0.003 [−0.09, 0.09]		0.07 [−0.03, 0.17]
Mindfulness (FFMQ)
Observe	−0.02 [−0.11, 0.07]	0.03 [−0.03, 0.07]	–		–
Describe	0.09* [0.001, 0.17]	−0.001 [−0.06, 0.05]	–		–
Actaware	−0.14* [−0.26, −0.02]	0.16 [0.09, 0.22]	–		–
Nonjudge	−0.03 [−0.11, 0.05]	0.001 [−0.06, 0.06]	–		–
Nonreact	0.03 [−0.11, 0.16]	−0.12 [−0.21, −0.02]	–		–
Cross-level interactions
Creative achievement*control	–	–	0.01 [−0.004, 0.02]		−0.01 [−0.03, 0.002]
Originality*control	–	–	−0.27 [−0.86, 0.13]		0.18 [−0.63, 0.98]
Fluency*control	–	–	−0.0003 [−0.07, 0.08]		0.01 [−0.06, 0.10]
Creative production*control	–	–	0.01 [−0.07, 0.09]		0.01 [−0.08, 0.11]
Random effects
Intercept (σ^200)
Without Level-2 predictors	0.13 [0.05, 0.37]	0.22*** [0.11, 0.45]		0.20*** [0.10, 0.40]	0.28*** [0.14, 0.55]
With Level-2 predictors	0.12 [0.03, 0.47]	0.10 [0.03, 0.37]		0.10 [0.03, 0.29]	0.21*** [0.09, 0.48]
Slope of momentary control (σ^21j)
Without Level-2 predictors	–	–		0.03 [0.003, 0.23]	0.05 [0.01, 0.22]
With Level-2 predictors	–	–		0.04 [0.01, 0.25]	0.06 [0.01, 0.24]
−2*log likelihood	4515.39	1229.51		1185.62	1294.18
R^2	.10	.20		.34	.26

MWD = Mind Wandering Deliberate scale; MWS = Mind Wandering Spontaneous scale; CAG = Creative Achievement Questionnaire; AUT = Alternative Uses Task; ADT = Alien Drawing Test; FFMQ = Five Facet Mindfulness Questionnaire; R² = proportion of variance attributable to the fixed and random effects. Numbers are parameter estimates and 95% BCa confidence intervals (in brackets; asymptotic confidence intervals for all random effects and the parameter estimates in the model predicting mind wandering). Significant (p < .05) parameters, according to the 95% BCa confidence intervals or likelihood ratio (LR) tests (random effects of the MLMs with continuous outcomes), are printed boldface. Asterisks pertain to the results of the Wald-type tests and the LR tests.

*p < .05, **p < .01, ***p < .001.

Perceived control (RH 1b)

Clustering was present at Level 2 for ratings of perceived control (ICC = .15; significant σ²₀₀ without the Level-2 predictors, see Table 2). Higher acting with awareness predicted more momentary control over mind wandering (b* = 0.35 [0.20, 0.49]), and higher nonreacting to inner experience (b* = −0.25 [−0.43, −0.04]) and creative production (b* = −0.14 [−0.27, −0.02]) predicted less momentary control (Table 2). Men reported higher control over mind wandering than women (d = −0.82).

Perceived usefulness (RH 2a and 3a)

Clustering was present at Level 2 for ratings of perceived usefulness (ICC = .11; significant σ²₀₀ without Level-2 predictors, see Table 2). Higher creative achievement predicted reporting more useful thoughts during mind wandering (b* = 0.20 [0.14, 0.27]). Also, higher usefulness was associated with higher momentary mood (b* = 0.10 [0.03, 0.16]), perceived novelty (b* = 0.12 [0.04, 0.19]), and, most strongly, control (b* = 0.34 [0.26, 0.42]; Table 2). More habitual spontaneous mind wandering predicted less useful thoughts during mind wandering (b* = −0.14 [−0.21, −0.07]). Random variation of the slope of momentary control was not indicated by the LR test. Also, none of the cross-level interactions of creativity with momentary control were significant.

Perceived novelty (RH 2b and 3b)

Clustering was present at Level 2 for ratings of perceived novelty (ICC = .15; significant σ²₀₀ without the Level-2 predictors, see Table 2). Less habitual spontaneous mind wandering (b* = −0.17 [−0.30, −0.03]), and lower originality (b* = −0.19 [−0.28, −0.10]) predicted more novel thoughts during mind wandering; higher novelty was further associated with higher ratings of perceived usefulness (b* = 0.14 [0.03, 0.24]; Table 2). Again, random variation of the slope of momentary control was not indicated by the LR test (Table 2). None of the investigated cross-level interactions of creativity with momentary control were significant either.

Discussion

This study investigated the association of control on novelty and usefulness of thought content during mind wandering in everyday life, controlling for the facets of mindfulness, different aspects of creativity (both self-reported and assessed with objective methods), and momentary mood alike. Utilizing an experience sampling approach, intensive longitudinal measurements were taken from each participant each day over a one-week period. The obtained nuanced pattern of results suggests that not only the distinction of deliberate and spontaneous mind wandering appears warranted, but also detailed looks at the facets of mindfulness and different aspects of creativity. The results highlight that neither mindfulness nor creativity should be treated as uniform constructs in this line of research and that correlates of mind wandering may differ in the lab and in daily life.

Higher acting with awareness was found to predict less mind wandering, but higher describing more mind wandering (discussed in more detail below). Hence, RH 1a was only partially supported. However, these findings are compatible with prior related observations (Agnoli, Vanucci, Pelagatti, & Corazza, 2018) that already highlighted the importance of these two facets for creative achievement and originality in moderation analyses with deliberate mind wandering. While the present study did not aim to replicate these interaction effects, it provided evidence that acting with awareness (and describe) may predict the overall likelihood of mind wandering. In contrast, the habitual deliberate and spontaneous mind wandering scales did not predict the actual frequency of mind wandering in everyday life in the present study. This highlights the utility and value of the applied ESM approach in the present study and mirrors prior results, which highlighted that correlates of mind wandering in the lab may differ from correlates in daily life (e.g., Kane et al., 2017).

Higher acting with awareness predicted also higher momentary control over mind wandering (RH 1b). However, nonreactivity to inner experience predicted lower momentary control. Thus, RH 1b was also only partially supported. However, concerning the contents of acting with awareness and nonreactivity to inner experience, directionally opposed associations with momentary control appear perfectly plausible. Whereas the former implies the deliberate enactment of (conscious) control, the latter implies the loosening of control (“letting it happen”). On the person level, these two mindfulness facets were both negatively associated with habitual spontaneous mind wandering (as in Seli, Carriere, & Smilek, 2014), which, itself, did not predict momentary control.

These findings thus highlight that the associations of mindfulness facets with momentary and self-rated habitual control of thought processes need not align, and need not even align for momentary control itself, as these two facets of mindfulness predicted different behavioral tendencies in everyday life. This, again, mirrors prior results on differences in the correlates of mind wandering in the lab and daily life (e.g., Kane et al., 2017), and further corroborate the notion that, optimally, mindfulness should not be treated as a uniform construct in research along these lines.

More momentary control was associated with more useful, but not more novel, thought content; less spontaneous habitual mind wandering predicted both usefulness and novelty. Thus, RH 2a was fully supported, whereas RH 2b partially. These results provide further evidence for the need to distinguish between controlled and uncontrolled forms of mind wandering and suggest that self-assessed habitual uncontrolled mind wandering is at least sufficiently valid and accurate to predict some aspects relevant to mind wandering and creativity, even though not the frequency of mind wandering itself (see above). As controlled mind wandering had no similar predictive effects, this suggests that only the relative lack of spontaneous mind wandering is relevant for the usefulness and novelty of thoughts.

There were no moderating relationships between creativity and momentary control on perceived usefulness and novelty of thoughts. Hence, RH 3a and 3b were not supported. Yet, higher creative achievement predicted higher usefulness of thoughts with a main effect, lending at least some support to the model of Fox and Beaty (2019). However, other main effects of the investigated creativity variables were less supportive of Fox and Beaty’s (2019) model: Higher originality predicted less mind wandering and less novel thoughts, and higher creative production predicted lower momentary control. While all these main effects were rather small in magnitude, this suggests that different aspects of creativity were differentially predictive of mind wandering and the usefulness and novelty of thoughts. This adds to the growing evidence that different measures and aspects of creativity may not fully agree with one another (see Said-Metwaly, Van den Noortgate, & Kyndt, 2017), which is exacerbated by the fact that, currently, no consensus definition of creativity and no universally agreed-upon instruments for its measurement exist. The present observations may be limited to specific aspects of creativity, the specific instruments for measuring them, or even the utilized scoring methods. Thus, the present results may serve as a starting point for future inquiries and need to be followed-up.

Consistent with previous evidence (Franklin et al., 2013), better mood during mind wandering was associated with more useful (but not more novel) thoughts. Lower mood was also associated with more frequent mind wandering, directly replicating findings of Killingsworth and Gilbert (2010). Lastly, women reported, with a large effect size, less control over thoughts during mind wandering in the current study than men. Gender effects apparently are a neglected topic in this field of research. It is well-known that women seem to be more prone to rumination than men (Johnson & Whisman, 2013). Inter alia, beliefs about emotion control, feelings of responsibility for the emotional coloring in relationships, and diminished control over negative emotions have been identified as underlying this effect (e.g., Nolen-Hoeksema & Jackson, 2001). Overall, gender differences in the construct domain of mind wandering need further inquiry.

Piecing together the pattern of results, it seems that mind wandering in the present sample likely had overall a negative (and possibly ruminative), rather than a positive and creative, tone (which could have been further fueled by the COVID-19 pandemic during the time of this study, see below). As the content of mind wandering itself was not measured, the present data provided no direct evidence for this. However, this interpretation would fit with the large gender difference in momentary control, the association of lower mood with more frequent mind wandering, and its differential associations with the mindfulness facets.

Mindfulness can be considered to be composed of the two components self-regulated attention (i.e., aspects of increased capacity of attention regulation) and orientation to experience (i.e., the monitoring of the stream of consciousness, characterized by an accepting and curious attitude; Bishop et al., 2004). Previous research has shown that the facets of the FFMQ can be assigned to these two components, with acting with awareness loading on orientation to experience, which has been specifically linked to beneficial effects of mindfulness on mental health in previous research (Burzler, Voracek, Hos, & Tran, 2019; Tran et al., 2014; Tran, Glück, & Nader, 2013). If the contents of mind wandering were mostly ruminative, this would explain the association of acting with awareness with fewer mind wandering and with more momentary control in the present study. On the other hand, describing has a loading on self-regulated attention, which, among non-meditators, has been, weakly, but still so, linked to symptoms of anxiety (e.g., Burzler, Voracek, Hos, & Tran, 2019; Tran, Glück, & Nader, 2013). As control, usefulness, and novelty were all interrelated in the present study, acting with awareness may thus be an important facilitator of creativity and creative processes (while at the same time protecting against rumination), which may explain the overall positive associations of mindfulness with creativity (Lebuda, Zabelina, & Karwowski, 2016).

The observed lack of association of momentary control with novelty could further indicate ruminative processes on the one hand, but also ordinary processes of problem-solving (Ruby, Smallwood, Sackur, & Singer, 2013) on the other hand. These two processes were not differentiated in the present study, but are characterized by the repetitive occurrence of the same thought contents (uncontrolled and maladaptive in rumination, controlled and adaptive in problem-solving), which diminish their subjective novelty. Future research may benefit from scrutinizing such processes more directly and assess them via the ESM, alongside the usefulness and novelty of thoughts.

Strengths and limitations

A major strength of the present study is its design, which allowed to investigate mind wandering with high external validity in everyday life, unbiased by memory or reporting effects. Furthermore, three indicators of creativity were utilized and assessed, each of which captured different aspects of creativity, with this triangulation adequately accounting for the multidimensionality of the construct. Another study strength is the high response rate of participants, which compares well to other studies in this field of research (e.g., Franklin et al., 2013; Poerio et al., 2013) and testifies to a conscientious attitude on the part of respondents. As well, analysis differentiated between the distinct levels of the data structure by consequently utilizing multilevel modeling.

The data collection took place during COVID-19 pandemic restrictions (World Health Organisation [WHO], 2020), which can be considered a serendipitous opportunity. This makes the present data unique with respect to the phase of data collection and opens up ample opportunities for replication studies. During the data collection period, exit and contact restrictions were in place for the general population in Austria (Müller, 2020) and Germany, which drastically changed the everyday life of many persons. Some studies have noted drastic effects of the pandemic on mood (Li, Wang, Xue, Zhao, & Zhu, 2020). Prevalence rates of depressive symptoms, anxiety symptoms, and sleep disorders have also significantly increased in Austria (Budimir, Pieh, Dale, & Probst, 2021). Accordingly, the participants in the present study reported considerable effects of the COVID-19 pandemic containment restrictions on their daily lives as well. Additionally, the magnitude of reported effects was associated with habitual mind wandering (both deliberate and spontaneous), average mood, and the acting with awareness FFMQ facet. The COVID-19 restrictions thus likely affected the results of this study in important ways. Future replication studies, conducted during times without pandemic restrictions, may assess the full scope of such effects (i.e., whether they only introduced differences in means or also altered the associations between variables).

The generalizability of results is limited by the rather homogeneous, well-educated, and relatively young participant pool. This segment of the general population is particularly adept at using smartphones, which may also have been a reason for the high response rate. Sample size at Level 2 was also comparatively low, which may have resulted in low analytical power for some of the investigated associations. Thus, the present results do not rule out that the predictions of Fox and Beaty’s (2019)model could indeed be correct. However, the present study suggests that any of the predicted interactions can be only small and could thus be likely less relevant overall. Also, some measures may have been confounded with sociodemographic or unmeasured third variables. For example, the CAQ may capture opportunity available to individuals as well as various personality traits independent of creative ability. Future studies should strive to include older persons, because it is already known that mind wandering decreases with age (McVay, Meier, Touron, & Kane, 2013), investigate also larger samples, and examine potentially confounding effects of third variables in more detail.

Another limitation may concern technical details in conducting this study. It was not feasible to verify how reliable the mail application notifications actually appeared on participants’ cell phone displays. Yet, if participants adhered to the trained assessment procedure in the event of a delayed notification, no biases should have occurred. Further, the present experience-sampling approach required participants to recollect whether their minds had actually wandered when they received the signal. This may have introduced some memory biases in some cases. Finally, the agreement rate concerning symmetry in the ADT was low, but apparently typical. This and the utilized scoring scheme for Originality in the AUT (for alternative scoring schemes, see Reiter-Palmon, Forthmann, & Barbot, 2019) may have also impacted the results.

Directions for future research

The results of the present study provide further evidence for the need to distinguish controlled and uncontrolled mind wandering. Future research should thus continue to consider the aspect of control during mind wandering. However, replication studies during times without pandemic restrictions are needed to fully determine the ways these restrictions have acted on the present results.

The definition of mind wandering needs to be further clarified in the future, as well as the different processes which may take place during mind wandering (Christoff et al., 2018). For example, it would be interesting to determine whether thoughts rated as less novel are related to problem-solving processes, rumination, or further thought processes. Investigating situational factors would be further interesting, as the frequency and nature of mind wandering appears to vary between individuals, but also within individuals. Overall, differences between measurements in the lab and daily life (e.g., Kane et al., 2017) need more consideration. Physiological measurements in combination with ESM in everyday life could provide further insights both into mind wandering and mindfulness (for a critical account and alternative avenues to assessment, see Lutz, Jha, Dunne, & Saron, 2015), as there is mounting evidence (Fischer, Voracek, & Tran, 2023; Tran et al., 2022) that self-reported mindfulness may lack validity and utility.

Qualitative approaches to the phenomenon of mind wandering, as used early on (e.g., Singer & McCraven, 1961), have almost completely disappeared from current research on this topic. However, qualitative approaches may be specifically promising concerning the definitional problems surrounding the construct of mind wandering and for identifying and differentiating the processes involved therein. At the same time, imaging techniques would be useful for pinpointing the neural correlates and functional brain mechanisms of mind wandering (Fox & Beaty, 2019). Using a diversity of methodological approaches to study mind wandering appears to be essential for research progress on this theme.

Past research has almost exclusively focused on well-educated and healthy young adults. Research accounting for all educational levels, as well as on non-adults and the elderly, is still needed. Also, focused investigations of extreme groups (e.g., extraordinary creative persons) could prove to be informative.

In conclusion, while mind wandering appears to be related to creativity, the crucial question remains under what circumstances mind wandering enhances creativity. Mood, as well as control over extent and content of mind wandering, and mindfulness all seem to play a role. Future research should specifically target the processes involved in mind wandering, as these may further clarify its adaptive and maladaptive functions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study and the SPSS code to reproduce the main analysis are openly available in OSF at https://osf.io/mvy3n/.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/10400419.2024.2326339