Everyday bodily movement is associated with creativity independently from active positive affect: a Bayesian mediation analysis approach

Previous (predominantly) laboratory studies reported positive relations of physical activity (or everyday bodily movement) with executive functioning, some even showed effects on creative thinking. Furthermore, positive-activated affect was found to be positively associated with everyday bodily movements and creativity. The mechanisms, however, underlying these relationships are poorly understood. The aim of this study was twofold: Firstly, we investigated whether everyday bodily movement was associated with creative performance. Secondly, we examined if positive-activated affect may mediate the association between bodily movement and creative performance. In a sample of 79 participants everyday bodily movement was recorded during five consecutive days using accelerometers. Creativity in the figural and verbal domain was assessed with performance tests, along with self-reported positive-activated affect as a trait. Findings revealed that creativity, positive-activated affect, and everyday bodily movement were associated with each other. However, positive-activated affect did not mediate the association between everyday bodily movement and creative performance. The pattern of findings argues for shared variance between bodily movement and creativity (fluency and originality) that is largely independent from variations in positive-activated affect.

Mediation effect of PAA on the association between bodily movement and total creative performance. In line with the Pearson correlations, the Bayesian mediation analysis revealed that bodily movement was positively associated with both creative performance (b = 0.27, lower CI = 0.06, upper CI = 0.49) and PAA (b = 0.23, lower CI = 0.01, upper CI = 0.46). These findings are depicted in Table 1.
The total effect of the mediation model was b = 0.32 (lower CI = 0.10, upper CI = 0.54) and the indirect effect was b = 0.04 (lower CI = -0.02, upper CI = 0.13). The indirect effect added only 13.51% (lower CI = -12.85%, upper CI = 39.87%) to the direct effect of bodily movement on creative performance (b = 0.27, lower CI = 0.06, upper CI = 0.49). This argues against a strong mediating effect of PAA on the relationship between bodily movement and creativity, since the CIs include a negative percentage. Due to the absence of a clear mediation, the pattern of findings indicates that activity and PAA are independently linked with creative performance (see Table 1). These results remained virtually unchanged when including age, gender, and BMI as covariates.
Additional analyses. Proportion of time spent with different types of bodily movement. In order to evaluate the association of physical activity with creative performance and PAA in more detail, exploratory correlation analyses were calculated with the proportion of time spent with bodily movements of four different intensity levels. As illustrated in Fig. 1, total creative performance showed significant positive associations with time spent in moderate (e.g., walking) and a negative association with movements of no to light intensity (e.g., sitting and lying). However, PAA was associated with time spent with very vigorous physical activity, but not with the proportion of time spent with sedentary behaviour or movement at low and moderate intensities. This pattern of findings indicate that the variance of CPM associated with PAA might be qualified by very vigorous activity. www.nature.com/scientificreports/ Originality and fluency of creative performance. Since the measure of total creative performance has been criticized due to its high correlation with the fluency of ideas 57-59 , we also investigated, which aspects of creativity (originality, fluency) were linked with bodily movement. The regression analysis showed that both originality (sr = 0.23, p = 0.038) and fluency (sr = 0.25, p = 0.024) were positively and independently from each other associated with physical activity. The significant regression analysis indicated that the quantity and the quality of ideas together shared 9.6% of variance with physical activity (F(2,76) = 4.06, p = 0.021). A similar regression analysis investigating the association of fluency and originality with PAA slightly failed to show significance (F(2,76) = 3.00, p = 0.056). Only fluency (sr = 0.27, p = 0.018), but not originality (sr = 0.10, p = 0.354) was associated with the PAA. Originality (r = 0.52, p < 0.001) and fluency (r = 0.70, p < 0.001) were correlated with total creative performance.

Discussion
This study examined firstly, whether everyday bodily movement is associated with creativity and secondly, if this association is mediated by participants' PAA. Importantly, this study replicated the reported positive relationship between bodily movement and creativity, which to date was predominantly found in laboratory studies 21,25 . We assessed bodily movement in everyday life by means of acceleration sensors. In contrast to former methods, the applied ecological assessment methodology objectively measures the level of spontaneous bodily movement in daily life-free from systematic errors of subjective self-reports 38 . A more detailed examination of bodily movement (i.e., categorization into different intensity levels) indicated that not only time spent with activities of high intensity, like sport and bouts of exercise, go along with increased creative performance 21,28,30 , but also everyday physical activities of a moderate intensity level 19 . The negative relationship between the proportion of time spent with light intensities-including sedentary behavior-and  www.nature.com/scientificreports/ creativity further underlines the a priori assumption that physical activity and bodily movement have beneficial effects on creative cognition 19 . This is also nicely in accordance with the anecdotal view that creative people use bodily movement to overcome mental blocks and lacks of inspiration 20,32,60,61 (but see 62 ). Bodily movement was also linked to PAA 40,41 , which in turn was associated with creativity 35,47,55 . Importantly and critically, while all three variables of interest were interrelated, the Bayesian mediation analysis indicated that PAA did not explain the relationship between bodily movement and creativity. This finding corresponds to a former study. Steinberg et al. 20 reported an impact of physical activity on creative performance and positive affect. However, they did not find evidence for a mediation. Importantly, together with Steinberg et al. 20 , the present finding suggests an additive effect of PAA and physical activity on creativity. This conclusion may also have important consequences: (1) Other psychological mechanisms have to be considered responsible for the observed relationship between physical activity and creativity. It might be hypothesized that the personality trait openness to experiences may serve as a potentially mediating variable, because it goes along with creativity and physical activity 63,64 . (2) To promote creativity, the induction of PAA plus the engagement in physical activity separately might be more fruitful in contrast to one strategy alone. This is in line with the observation that PAA was mainly associated with the fluency of ideas, but bodily movement seemed to be linked with the quantity and quality of ideas.
In accordance with this, Curnow and Turner 39 reported an additional effect in a music (i.e., arousal) and exercise combined group compared to exercise and music alone groups in TTCT fluency, but they did not find differences in originality. However, this study investigated acute effects of exercise and emotional arousal on creativity, while the present study focused on the chronic aspect of this relationship, thus hampering a direct comparison of results. To overcome this limitation of solely focusing on chronic or acute mechanisms in future studies, experience-sampling methods (i.e., ecological momentary assessment of everyday life behavior), which enable the measurement of creativity, affective states, and bodily movement in the very same moment of time, should be applied 44,65 . This innovative approach may allow a more detailed insight into the mechanisms responsible for the relationship between physical activity and creativity. In conjointly using ecological momentary assessment and multi-level analyses, the chronic as well as acute effects of physical activity (and positive affect) on creative performance may be distinguished.
While in accordance with previous research higher PAA was associated with more bodily movement 41,44-46 the more detailed analyses-using the proportion of time spent with bodily movement at different intensity levels-showed a divergent pattern of correlations. PAA was linked with very vigorous physical activity, but not with moderate and low intensities 40,42 . These additional analyses indicate one further explanation for the absence of a mediation effect. While a sedentary lifestyle seems to be associated with lower creativity scores, it shows no association with the affect. With other words: Because affect and creativity are associated with different intensity levels of everyday bodily movement, movement associated changes in affect might therefore be independent from movement associated changes in creativity. This conclusion is well in line with an additive effect of affect and physical activity on creativity. Despite the promises of ecological momentary assessment studies as outlined above, the present study was interested to assess the trait component of all variables and tested whether variance of the proposed mediator (i.e., PAA) overlaps with the shared variance of the predictor (i.e., bodily movement) and the criterion (i.e., creative performance).
This study is not without limitations. It is important to note that-due to the cross-sectional study design-the reported correlations are not causal in nature. Hence, it would also be possible, for instance, that creativity leads to more positive affect 65 and higher levels of positive affect result in more everyday bodily movement 44 . Due to the moderate (and even low) reliability of the measures of bodily movement, PAA, and creativity, results should be treated with caution. However, physical activity was assessed on five different days including the weekend. Thus, a moderate internal consistency is well in line with the assumed variations of bodily movement throughout a week 66 . Similarly, total creative performance was assessed by two measures in two domains (i.e., figural and verbal), further decreasing internal consistency, which was in the reported range of various scoring methods in this field 59 . Although, the total creative performance score showed the expected association with bodily movements, it has also been criticized [57][58][59]67 , and a differentiation between quality (i.e., originality) and quantity (i.e., fluency) of ideas has been suggested. In accordance with the total creative performance score, originality and fluency were positively associated with bodily movement, however, only fluency was associated with PAA. This is in line with many laboratory studies investigating the association with physical activity by means of creativity measures combining fluency and originality of ideas 19,28,30 . Finally, the sample size of the present study might be too small to detect a small effect of PAA on the association between creativity and physical activity by means of a frequentist approach. However, the use of informative priors in a Bayesian mediation analysis increases the power of analyses and the precision of estimates 56 . Nevertheless, and not at least due to the high number of correlations in this study, a replication of the findings, also by means of a momentary ecological assessment approach is recommended before far-reaching conclusions are warrantable.
To sum up, the present findings are in accordance with the common notion that physical activity has many beneficial effects. Besides the well-known effects on physical health, a greater degree of everyday bodily movement is associated with higher levels of PAA (and subjective well-being 40 ) and better creative cognition 12,19 . This study represents a first step to investigate this complex relationship in everyday life. The findings argue for an association between everyday bodily movement and creativity that is not merely attributed to the PAA. Further studies are certainly warranted to elucidate other pathways. creativity tasks. Verbal creativity. In the Alternate Uses (AU) task 22 participants had to produce as many original and creative uses of conventional everyday objects (e.g., umbrella, car tire) as possible within 1 min each. In total participants worked on five items. This procedure resulted in 1,164 separable ideas in total. The originality of each idea was rated by three instructed and independent raters on a four-point rating scale ranging from 1 (not original at all) to 4 (highly original; cf. Consensual Assessment Technique 68 ; see, e.g., Rominger et al. 69 ). The inter-rater agreement was ICC (  Total creative performance. In order to indicate participant's total creative performance, the z-transformed verbal and figural creative performance scores were aggregated (α = 0.45). For additional analyses the fluency and originality scores were calculated by the sum of the z-transformed scores. Cronbach's alpha for fluency was α = 0.63 and α = 0.59 for originality.
Objective measurement of everyday bodily movement. Bodily movement was recorded by means of tri-axial acceleration sensors (ActiGraph®, Model GT3X + , weight of 19 g). All participants carried the sensor on the right side of their hip, attached with an elastic belt 66,73,74 . This approach provides a measure of metabolically relevant whole-body movements and their intensity levels 66 . The acceleration data were sampled with 30 Hz and activity counts were analyzed using 1 min epochs (i.e., counts/min, CPM; see Schwerdtfeger et al. 44 for a similar approach). The activity counts were the composite vector magnitude of the three axes 75,76 . The CPM as the main output of the ActiGraph 77 was validated in a high number of studies (e.g., 78 ; for details how ActiGraph counts are generated, see Brønd, Andersen, & Arvidsson 79 ). All data were analyzed with the software ActiLife (Version 6.13.2). The acceleration data were recorded on five consecutive days during waking hours, in order to reliably and representatively measure the degree of participants' everyday bodily movement 66,80 . The recordings started on Wednesday 00:00 and ended on Sunday at 23:59. Since all participants were instructed to note the sensors' nonwearing times in a calendar template (e.g., during bathing, or sleeping), we were able to exclude invalid periods of sensor data (i.e., counts of zero) from CPM estimation. The minimum of wearing time per day was set to 600 min (M = 839.89 min, SD = 79.69 min) 38 . Non-activities longer than 90 min during the self-rated wearing time were treated as invalid non-activities and all counts of zero during this time were excluded from the CPM estimation 81 . The resulting data were finally quality checked for spurious counts (i.e., CPMs above 20,000 were inspected in more detail) 80 www.nature.com/scientificreports/ feel in general on all 20 adjectives of the PANAS on a five-point Likert scale (from 1 = not at all to 5 = very much so). Similar to Schwerdtfeger and Gerteis 86 , PAA was indexed by the mean of the four items 87 ("active", "alert", "attentive", "inspired"; M = 3.47, SD = 0.55, Cronbach's α = 0.61).
procedure. All participants were tested separately. Firstly, they signed the informed consent, followed by the PANAS and the divergent thinking tests (AUT, TTCT). Participants were familiarized with the study protocol and the technical equipment (acceleration sensor, belt). Following the instruction, the accelerometer was initialized and attached. Finally, participants were instructed to report the sensor wearing times in a template. After the objective measurement of habitual bodily movement, participants returned the sensors and finally rated the typicality of the recording days on a four-point rating scale (1 = absolutely typical, 2 = rather typical, 3 = rather untypical, 4 = totally untypical). No participant rated these days as totally untypical (17.72% absolutely typical, 69.62% rather typical, 12.66% rather untypical).

Statistical analysis.
In order to answer the first study question, Pearson correlation was used to test the relationship between everyday bodily movement and total creative performance. Furthermore, as an important prerequisite for a mediation, the inter-correlations of all three variables of interest (i.e., CPM, PAA, and total creative performance) were calculated 88 . Bayesian mediation analysis was applied to evaluate if the association between everyday bodily movement and creativity was due to the influence of PAA. The statistical program R (version 3.4.3) 89 with the package brms (version 2.7.0) 90 was used. A Bayesian approach instead of a common frequentist approach was applied, because of (1) the moderate sample size of 79 participants 56,91 and (2) the a priori knowledge about the intercorrelations. The use of prior information in Bayesian statistics increases the power of the applied mediation analysis 56 . Firstly, physical activity is associated with increased creativity 19,21 . Secondly, physical activity is positively associated with PAA 44 , and, thirdly, PAA is linked with creativity 47,55 . Based on relevant literature, normally distributed informative priors for the regression coefficients of bodily movement and creativity (µ = 0.12, σ 2 = 0.10; Etnier et al. 12 reporting a d of 0.25), bodily movement and PAA (µ = 0.27, σ 2 = 0.10; Reed & Buck 40 reporting a d of 0.57), and creativity and PAA (µ = 0.08, σ 2 = 0.10) 55 were used 56 .
Samples were derived by Markov Chain Monte Carlo (MCMC) algorithm, 4 chains with 5,000 iterations (1,000 warm-up samples for each chain) and 16,000 post-warmup samples were used. Unstandardized effect estimates (b), 95% credible intervals (CI), and the potential scale reduction factor on split chains (Rhat) are reported. For Rhat, values close to 1 indicate that the chains converged (i.e., the number of iterations was sufficient). To evaluate the mediating role of PAA on the association between bodily movement and creativity the direct, indirect, and total effect, and the effect proportion of the mediation are reported. This was done with the R package mediation (version 4.4.6) 92 . The mediating effect of PAA is considered as present, when the credible intervals (CIs) of the indirect effect are positive and do not include zero.
In additional analyses the association between the intensity levels of physical activity and total creative performance with the proportion of time spent with physical activities in different intensity levels were calculated. Equivalent correlations were calculated for PAA. Since the total creative performance score has been criticized to be mainly driven by the fluency of ideas 58,93 , additional analyses were calculated to investigate the association of originality and fluency with bodily movements, by means of a regression analysis. This approach allows to investigate the unique variance quality and quantity of ideas share with physical activity. A similar regression analysis was calculated for PAA.

Data availability
The data will be available from the first author. www.nature.com/scientificreports/