Right-handed participants in this study were recruited from individuals majoring in dance (DANCE) and non-dancer controls (CON). After 14 participants chose to withdraw from the study and an additional 12 were excluded due to structural brain abnormalities, severe motion artifacts, or technical issues with data collection, a total of 29 DANCEs (mean age 23.1 ± 2.9 years) and 28 CONs (mean age 22.8 ± 1.6 years), carefully matched for age and education level, were included in the analyses. None of the participants in both the DANCE and CON groups reported having received any training in sports. All participants were selected from the identical sample previously detailed in our earlier study, and specifics regarding demographics and dance training can be found in Table 1 of the published work (Yang et al., 2023). The study received approval from the Institutional Review Board of Taipei Veterans General Hospital, and written informed consent was obtained from each participant.

The Abbreviated Torrance Test for Adults (ATTA) was employed to assess general creativity (Chen, 2006). The ATTA battery includes one verbal and two figural tests, with four norm-referenced creativity indicators (fluency, originality, elaboration, flexibility), a creativity index (the sum of the aforementioned 4 measures), and two criterion-referenced creativity indicators (verbal and visual creativity) calculated for an overall creativity profile of each participant (Kharkhurin and Samadpour Motalleebi, 2008; Althuizen et al., 2010; Kharkhurin, 2010; Shen and Lai, 2014; Sunavsky and Poppenk, 2020). A comparison of the creativity profiles between the DANCE and CON groups was conducted based on the six indicators of general creativity. Between-group differences were assessed using a two-sample t-test (SPSS Statistics version 27.0, SPSS Inc., USA), with statistical significance set at p < 0.05.

Magnetic resonance imaging was conducted using the 3T MAGNETOM Trio™ system, with participants positioned supine within the scanner. To minimize motion artifacts, foam cushions were used for head fixation inside the head coil. Resting-state functional scans were obtained through a T2*-weighted gradient echo planar imaging (EPI) sequence with the following parameters: repetition time (TR) = 2500 ms, echo time (TE) = 30 ms, flip angle = 90°, field of view (FOV) = 220 × 220 mm², slice thickness = 3.4 mm, slice number = 40, matrix size = 64 × 64, tilted angle = 30°, and voxel size = 3.4 mm × 3.4 mm × 3.4 mm. Each resting-state fMRI time series consisted of 200 volumes, with a duration of 500 s per time series. Additionally, T1-weighted structural images were acquired using the magnetization-prepared rapid gradient echo (MPRAGE) sequence with the following parameters: TR = 2530 ms, TE = 3.03 ms, flip angle = 7°, FOV = 224 × 256 mm², matrix size = 224 × 256, and slice thickness = 1 mm. Participants were instructed to maintain a motionless and alert state, keeping their eyes open and refraining from engaging in any specific thoughts.

The advanced DPARSF module V5.4 was used to preprocess the resting-state fMRI data (Yan et al., 2016). The preprocessing involved a series of sequential steps, starting with slice timing correction and followed by realignment to correct for head motion. Participants displaying head motion exceeding 2 mm displacement or 2° rotation in any cardinal direction were excluded. Subsequently, T1-weighted images were co-registered to the mean functional image using intra-subject spatial alignment. The segmentation of gray matter, white matter, and cerebrospinal fluid was carried out using the unified segmentation model. Nuisance regression utilized the Friston 24-parameter model (Friston et al., 1996) and default masks from SPM, eliminating head motion parameters and signals from white matter and cerebrospinal fluid. Spatial normalization to a study-specific DARTEL template (Ashburner, 2007), transformed to MNI space, was performed with image resampling to 3 mm isotropic voxels. Spatial smoothing was applied using a Gaussian kernel with a full width at half-maximum (FWHM) of 6 mm. Temporal band-pass filtering (0.01−0.1 Hz) was implemented to minimize high-frequency noise and low-frequency drift. Global signal regression (GSR) was not applied due to its tendency to amplify negative correlations and distort between-group differences (Murphy et al., 2009; Weissenbacher et al., 2009; Saad et al., 2012).

Metacognition-related regions, including the rostrolateral PFC (rlPFC, BA10), dorsolateral PFC (dlPFC, BA46), dACC/pre-SMA (BA32), medial PFC (mPFC, BA10/32), insula/IFG (BA47), precuneus (BA7/23), and ventral striatum, were defined as seed regions of interest (ROIs) since they have been identified in various tasks-based fMRI studies (Fleming et al., 2012; McCurdy et al., 2013; Morales et al., 2018; Vaccaro and Fleming, 2018). These seed ROIs were constructed as twelve 10-mm radius spheres centered at MNI coordinates identified by Morales et al. (2018) and Vaccaro and Fleming (2018) (see Table 1 for details of ROIs). The creation of these spheres was executed using WFU Pickatlas 3.0.5 (Maldjian et al., 2003). Given that dancers dynamically engage different aspects of metacognitive functioning for their learning and performance, it’s logical to merge individual ROIs into a unified, overarching composite ROI for resting-state functional connectivity (FC) analysis. This approach is rooted in the belief that these dispersed regions, having interconnected functions, are likely to function in a synergistic and holistic way (Rasero et al., 2018). The reference time course was derived by averaging the time courses of all voxels within this composite ROI consisting of 12 predefined ROIs. The FC map was then generated by assessing Pearson’s correlation coefficients (r) between the reference time course and the time course of each voxel of the brain. The r-value of each voxel was transformed to a z-value using Fisher’s r-to-z transformation to normalize the distribution. Multiple regression analyses were conducted on all z-transformed FC maps for controlling the effects of age and sex. Between-group comparisons were examined using two-sample t-tests on FC maps, with significance set at peak-level thresholds p < 0.005 and p < 0.001, followed by cluster-level p_FWE < 0.05 in SPM.

This study aimed to explore the impact of the interconnectedness between NCM and NCD on dancers’ general creativity performance, evaluated through the ATTA test battery. Drawing from the findings of May et al. (2020), three creativity indicators—fluency, originality, and flexibility—which exhibited a notable increase in dance students following metacognitive skills training were probed. Regions displaying significant between-group differences (DANCE vs. CON) in NCM-seeded FCs were identified. Spherical ROIs, each centered at the coordinates of these significant regions with a radius of 5 mm, were generated. The z-values extracted from these spherical ROIs were then correlated with ATTA metrics. Statistical significance was set at p < 0.05. Further, to address multiple comparisons, a Bonferroni correction was applied by adjusting the p-value to 0.0166 (0.05 divided by 3), given the three measures (fluency, originality, and flexibility) under examination.

The DANCE group exhibited significantly elevated originality scores on the ATTA (DANCE: 17.17 ± 1.77, CON: 15.32 ± 2.51, p = 0.002), with no discernible between-group differences observed for fluency, elaboration, flexibility, visual creativity, verbal creativity, or creativity index. These findings are derived from the identical sample and results reported in our earlier study (Yang et al., 2023).

The DANCE group demonstrated elevated interconnectedness between NCM and NCD. The targeted motor components of NCD included the bilateral putamen, bilateral globus pallidus (GP), left posterior cerebellum (lobule VI and crus I), right SMA, and right dACC/cingulate motor area (CMA). Moreover, these target regions also covered non-motor components of NCD, such as the bilateral anterior insula (AI), right IFG, left hippocampus, left STG, left mediodorsal thalamus, and left amygdala. Figure 1 and Table 2 provide additional details.

The DANCE group demonstrated significant negative correlations between originality scores and the strength of FCs linking NCM with NCD, specifically the left putamen (r = −0.529, p = 0.003) and left GP (r = −0.422, p = 0.023) (Figures 2A, B). On the contrary, the DANCE group displayed distinct positive correlations between flexibility scores and the strength of FCs linking NCM and NCD, specifically the left putamen (r = 0.416, p = 0.025), left GP (r = 0.494, p = 0.006), and left cerebellar crus I (r = 0.642, p < 0.001) (Figures 2A, B, D). The left AI, a common neural substrate of NCM and NCD, was also targeted (r = 0.54, p = 0.003) (Figure 2C). Notably, the CON group exhibited no significant correlations in these aspects.

Upon detailed examination, within both the DANCE and CON groups, no substantial relationships were identified between the strength of NCM-NCD FCs and other ATTA metrics. These parameters encompass the creativity index, fluency, elaboration, as well as verbal and visual creativity metrics.

In dancers, the NCM exhibit increased intrinsic FCs involving the AI, IFG, dACC/CMA, and rlPFC (Figure 1). Together with the mediodorsal thalamus, which is the target region of the extrinsic FC of NCM, all these regions collectively form the cingulo-opercular network, a key neural network involved in metacognition (Dosenbach et al., 2008; Morales et al., 2018). The cingulo-opercular network is recognized for its central role in the cognitive control of salience detection, reorientation, and mental switching, allowing for the flexible allocation of processing resources to other goal-relevant networks (i.e., the sensorimotor network) (Cocuzza et al., 2020). This neural network plays a significant role in fostering cognitive flexibility and participating in advanced cognitive functions such as attention, inhibition control, action preparation, working memory, and sensation (Camilleri et al., 2018). The AI and dACC/CMA have pivotal functions in detecting salience and integrating sensory, emotional, and cognitive information to foster self-awareness and social behavior (Craig, 2009; Menon and Uddin, 2010). The rlPFC empowers individuals to focus on environmental changes and on self-generated or sustained mental representations, often termed as “thoughts in our head” (Burgess et al., 2007). The enhanced intrinsic connectivity within the cingulo-opercular network may reflect the enhanced metacognitive abilities in dancers following extensive training.

In dance training, embodied metacognition demands that dancers comprehend dance concepts, infuse meaning into their movements, and apply their knowledge by deciding how to organize elements of body, gesture, locomotion, time, space, and energy (Hanna, 2014). Through extensive physical practice, rapid interactions are facilitated, allowing for quick adjustments in anticipation of performance outcomes. As expected, expert dancers demonstrated enhanced extrinsic FC between NCM and NCD (particularly motor components) as the neurological underpinning of (dancer) domain-specific embodied metacognition. This tight cognitive-dance movement interaction aligns with our earlier discoveries of optimized cortico-basal ganglia and cortico-cerebellar loops in dancers, particularly the interconnectedness of motor and cognitive/associative circuits (Yang et al., 2023). The specific regions targeted within the motor components of NCD encompassed both subcortical structures (putamen, GP, and posterior cerebellum) and cortical motor areas (SMA/CMA) (Figure 1). These motor substrates are acknowledged for their participation in executing movements (Haber, 2003; Errante and Fogassi, 2020). However, they play distinct roles in motor processing: the putamen regulates and facilitates voluntary movements, the GP inhibits movement, the posterior cerebellum coordinates movement and corrects prediction errors, and the SMA/CMA is involved in movement planning and anticipation (Hardwick et al., 2013; Koziol et al., 2014; Seghezzi et al., 2019; Rocha et al., 2023). The enhanced connectivity is consistent with the idea that dancers exercise heightened engagement in complex motor cognition, involving activities like nuanced action observation/simulation and refined motor imagery (Henschke and Pakan, 2023).

In dancers, the heightened FCs between NCM and NCD also involve the hippocampus, STG, and amygdala, as well as AI (the shared neural substrate of NCM and NCD) (Figure 1). These regions play roles in spatial cognition, rhythm synchronization, salience detection, and emotional processing. In the spatial dimension of dance, extensive training enhances dancers’ balance and spatial orientation skills, accompanied by observable increases in gray matter volumes in the hippocampus, insula, and CMA, setting dancers apart from non-dancers (Dordevic et al., 2018). Proficient dancers, in the temporal dimension of dance, exhibit heightened cortical thickness in the STG, particularly linked to rhythm synchronization, melody discrimination, and dance imitation, facilitating auditory–motor interaction in rhythmic contexts (Karpati et al., 2017). In the cognitive and emotional dimensions of dance, the AI serves a crucial integrative role, bridging physical, cognitive, and emotional domains, enabling advanced cognitive control and interoceptive awareness by converging various sensory and affective inputs (Craig, 2009, 2011). Both the AI and amygdala significantly contribute to processing socio-affective information, fostering emotional awareness and reactions associated with dancers’ empathic abilities (Gujing et al., 2019; Zardi et al., 2021). Alongside these two regions, the cerebellar crus I, in addition to participating in motor processing, plays a pivotal role in perception, emotion, and social cognition (Koziol et al., 2014; Baumann et al., 2015; Adamaszek et al., 2017; Van Overwalle et al., 2020). Overall, the increased connectivity may lead dancers to integrate top-down processes guided by knowledge with bottom-up processes driven by their dance experiences (Moffett, 2012).

The coordinated functioning of the identified areas, via both intrinsic and extrinsic connections in the NCM and NCD, could underpin the neural framework for dancers’ embodied metacognition. This coordination may heighten their metacognitive awareness and potentially improve their artistic expression in dance.

In this study, we substantiated the connections between NCM-NCD FCs and the general creativity performances in dancers, employing the ATTA. However, there are points for further consideration. Focusing on neuroplasticity in dancers and requiring group comparisons, we used a well-established psychometric creativity test more aligned with our goals, allowing us to examine creativity’s cross-domain effects in dancers. Since specialized experience, as seen in choreography and movement creativity, plays a role in both general and domain-specific creativity (Hong and Milgram, 2010; Qian et al., 2019), more detailed behavioral studies are needed to fully understand how domain-specific skills in dance interact with general creative abilities (Teng et al., 2021). This deeper exploration could reveal important insights into the complexities of creativity, both in dance and across various fields. Although the cross-sectional design may not be ideal for determining the specific duration and intensity of training required to manifest functional connectivity benefits for stimulating creative thinking, the findings could provide insights into the neurological basis for the positive effects of neuroplastic reorganization through dance. This non-pharmacological intervention may enhance motor and cognitive abilities in individuals with neurological diseases (Bek et al., 2022; Wu et al., 2022; Meulenberg et al., 2023).