This study recruited 27 healthy university students majoring in the visual arts (VA, mean age 24.0 ± 1.7, 5 men) and 27 healthy non-artists matched for age and education (control group: CON mean age 23.2 ± 1.6, 4 men). The creative mediums of students in the VA group included oil paints, ink, sculpture materials, and/or multimedia. The average duration of artistic training was 11.07 ± 4.6 years. Students in the CON group had no more than 3 years of institutional training in the visual arts. All participants self-reported right-handedness without metal implants, brain damage, or neuropsychiatric diseases. The Beck Depression Inventory (BDI) (Beck et al., 1996) and the Beck Anxiety Inventory (BAI) (Beck and Steer, 1990) were used to exclude participants with obvious emotional liability. We also used the Wechsler Abbreviated Scale of Intelligence (WASI-III) (Chen and Chen, 2002) to ensure correspondence between the two studied groups in terms of general intelligence. This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Taipei Veterans General Hospital with written informed consent obtained from all participants.

This study was part of a project on neuroaesthetics, which addressed issues pertaining to domain-general and domain-specific neural organization among art students in a variety of fields (visual arts, dance, piano, strings, vocals, percussions) and non-artist healthy controls. To facilitate inter-group comparisons, we had all of the artists undergo the same psychological assessments and neuroimaging using the same scanning protocols. In the current study, we focused exclusively on VAs and controls. Creativity can be a core mental competence of VA since the process of creating artwork reflects VA’s creativity (Getzels and Csikszentmihalyi, 2020). Thus, all participants took the self-reported 40-item Chinese version of the Abbreviated Torrance Test for Adults (ATTA) to assess their aptitude in tasks involving visual (figural) and verbal manipulation (Chen, 2006). The ATTA is commonly used for cross-artist group comparisons (not addressed in the current study); however, it includes a figural part that engages creative drawing mirroring the visual art training of VAs. The ATTA measures the ability to think creatively in terms of fluency, originality, elaboration, and flexibility (Chen, 2006). Fluency refers to the number of ideas that a participant can generate in a limited time. Originality indicates one’s ability to create unique ideas. Elaboration indicates the ability to embellish ideas with details. Flexibility indicates one’s ability to generate many different ideas (Althuizen et al., 2010; Shen and Lai, 2014). The ATTA creativity index (CI) score refers to the sum of the four capacity scores. We followed standard protocols in administering and scoring the tests (Chen, 2006). SPSS Statistics (v. 23.0, IBM Corp., Armonk, NY, USA) was used for all psychological evaluation analyses. The results of the psychological evaluations were considered significant at p < 0.05.

Resting-state fMRI data were acquired using a 3T MAGNETOM Trio™ (Siemens, Erlangen, Germany) at the National Yang-Ming University. During scanning, participants lay supine with their heads fixed using foam cushions to minimize head motion. Scanning was performed in a T2*-weighted echo-planar imaging (EPI) sequence with the following parameters: 40 axial slices, TR = 2500 ms, TE = 30 ms, flip angle = 90°, FOV = 220 mm × 220 mm, slice thickness = 3.4 mm, matrix size = 64 × 64, and voxel size = 3.4 mm × 3.4 mm × 3.4 mm. A total of 200 contiguous functional volumes were collected from each participant. High-resolution T1-weighted 3D structural images were acquired using a magnetization-prepared rapid acquired gradient echo sequence [MPRAGE; repetition time (TR)/echo time (TE) = 2530 ms/3.03 ms, flip angle = 70°, field of view (FOV) = 224 mm × 256 mm × 192 mm, in-plane matrix size = 224 × 256 × 192, in-plane resolution = 1 mm]. All subjects were instructed to relax, remain still with their eyes open, think of nothing, and refrain from moving or falling asleep. All participants received brief training on how to focus their attention through breathing before scanning began, and all participants maintained a similar state throughout the actual experiment.

Data preprocessing was performed using the Data Processing Assistant for Resting-State fMRI (DPARSF) V4.5 Advanced Edition (State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, China), which is based on the Data Processing and Analysis of Brain Imaging (DPABI) Toolbox version 4.1¹ (Yan et al., 2016), with statistical parametrical mapping 12 (SPM 12; Wellcome Trust Center for Neuroimaging, University College London, London, UK) in Matlab 2015b (MathWorks, Inc., Natick, MA, USA). Based on experience in previous studies, the preprocessing of functional images was performed as follows: (1) slice timing correction; (2) realignment of images to the mean volume for correction of head motion; (3) co-registration to map functional information of resting fMRI images into an anatomical space (T1-weighted images) via intra-subject spatial alignment; and (4) segmentation of gray matter, white matter (WM) and cerebrospinal fluid (CSF) from coregistered T1 images using the unified segmentation model (Wu et al., 2016). Subjects with any instances of head movement exceeding 2 mm or 2° were excluded from further processing. The following nuisance variables were regressed: (1) six parameters of head movement calculated based on head motion with the Friston 24-parameter model translation and rotation during realignment in SPM12 (Friston et al., 1996); (2) the mean signal within the lateral ventricles for cerebral spinal fluid; and (3) the mean signal within a deep white matter region (centrum ovale). The images were normalized to the custom template from T1 weighted images of all subjects developed by the Montreal Neurological Institute (MNI) with resampled voxels at 2 mm × 2 mm × 2 mm. The resulting time series in each voxel was then linearly detrended and bandpass filtered (0.01–0.1 Hz) to extract low-frequency oscillations. Global signal regression (GSR) was not performed as it has been shown to exaggerate negative correlations (Murphy et al., 2009; Weissenbacher et al., 2009) and/or to distort group differences (Saad et al., 2012). We used WFU Pick Atlas toolbox² to generate a visual system template based on the modified human visual pathway model by Choi et al. (2020). The visual system template includes the visual area [V1, V2, V3, V4, and V5/MT (BA 17, BA 18, and BA19)], inferior temporal area (BA 20), angular gyrus (BA 39), supramarginal gyrus (BA 40), and superior parietal lobule (BA 5, BA 7).

ReHo maps were computed using Kendall’s coefficient of concordance (KCC) of the time series between a given voxel and its nearest neighbors (26 voxels) in a voxel-wise manner (Zang et al., 2004). The ReHo map of each subject was divided by its own global mean and then spatially smoothed using a 3D Gaussian kernel with 6 mm full width at half maximum (FWHM). Comparisons between groups of ReHo maps masked by the visual system template were examined using a two-sample t-test in SPM. The peaks of significant clusters were then selected as ReHo-based seeds. Statistical significance was set at an uncorrected voxel level of p < 0.005, followed by the family wise error (FWE)-corrected cluster level of p < 0.05.

The preprocessing procedures were the same as those for the ReHo analysis, except for spatial smoothing, which was performed using a 6-mm FWHM Gaussian kernel prior to ReHo-seeded FC analysis. We observed significant between-group differences in terms of ReHo-seeded FC masked by the visual system template in all regions of interest (ROIs). Mean time-series activity was extracted within the spherical ReHo-seeded regions (5 mm radius) (Yan et al., 2013). ReHo-seeded FC was assessed between ROIs and the whole brain in a voxel-wise manner. Individual FC maps were then generated by computing the Pearson’s correlation coefficient (r) between the seeds and the related brain regions. After calculating the correlation between the reference time course and the time course of each voxel in the brain, the r-values were converted into z-values using Fisher’s r-to-z transformation to normalize the distribution. When analyzing differences between groups, two sample t-tests were performed on the ReHo-seeded FC maps of each seed with significance set at an uncorrected voxel level of p < 0.001, followed by an FWE corrected cluster level of p < 0.05 in SPM. Bonferroni corrections were made for multiple comparisons by adjusting the p-value divided by the number of seeds analyzed.

VAs can be considered skilled experts in creative production (Degarrod, 2016); therefore, we used SPSS statistical software (v. 23.0, IBM Corp., Armonk, NY, USA) to compare the VA and CON groups in terms of the correlation between ReHo and ReHo-seeded FC maps (masked by the visual system template) and ATTA scores. Training effects among VAs were revealed by correlating variables in visual arts training (duration of visual arts training, duration of daily practice hours, and average amount of practice time per week) with ReHo and ReHo-seeded FC maps, respectively. We extracted the z-scores of significant peaks from individual ReHo and ReHo-seeded FC maps to perform group comparisons and assess two-tailed correlations between the z-scores and ATTA scores and variables of visual art training, respectively. The significant level was thresholded at p < 0.05.

We observed no significant differences between the groups in terms of age, sex, level of education, depression, anxiety and intelligence. Compared to the control group, the VA group presented significantly higher scores for visual (figural) creativity (VA: 5.52 ± 2.06, CON: 3.03 ± 2.03, p < 0.001), fluency (VA: 16.1 ± 1.6, CON: 15.1 ± 1.7, p = 0.044), elaboration (VA: 17.5 ± 1.8, CON: 15.3 ± 2.4, p = 0.004), and flexibility (VA: 15.4 ± 1.6, CON: 14.5 ± 1.6, p = 0.041). They also presented a higher ATTA CI (sum of measurements of four categories of creative capacity; VA: 72.3 ± 6.1, CON: 65.0 ± 8.0, p < 0.001). Note that the VA group demonstrated also a sub-significant trend of higher originality performance as compared to the control group (VA: 17.0 ± 2.2, CON: 15.8 ± 2.6, p = 0.075) (Table 1).

To identify the cardinal functional hubs of the VAs, we quantified intraregional functional integration/segregation by calculating the voxel-wise ReHo value. The ReHo value was significantly higher in the VA group than in the CON group in the right SPL, right precuneus, left ITG, left ITG/fusiform gyrus (FG), left AG, and left middle occipital gyrus (MOG) of the visual system (Figure 1 and Table 2). These regions are associated with visual imagery (Ishai et al., 2000; Mechelli et al., 2004), visuospatial processing (Tres and Brucki, 2014), and the perception of objects, faces, and scenes (Sugase-Miyamoto et al., 2011; Conway, 2018). Notably, these regions are also engaged in elementary constructs of visual productivity (e.g., visual capacity, formation of higher-level complex visual representation, divergent thinking, long-term memory storage, and visual imagery) (Miyashita, 1993; Ishai et al., 2000; Mechelli et al., 2004; Sugase-Miyamoto et al., 2011; Zhang et al., 2011).

Six regions of interest (ROIs) identified in ReHo analysis were used as seeds for FC analysis. The VA group presented a higher FC of the left ITG-right cuneus, -right MOG, and -bilateral SOG (Figure 2 and Table 3). These targeted regions are associated with the functional processing of visual characteristics of objects and scenes, object recognition, and form representation.

Our hypothesis posited that the VA group would display distinct neurodynamics in the visual system compared to the CON group. We also expected the strength of FC to be associated with visual creativity score, based on prior research (Beaty et al., 2018). In our study, we found a significant positive correlation (r = 0.500, p = 0.002; Figure 3A) between the CI score and the strength of FC between the left ITG and the right SOG in the VA group. However, no significant correlation was observed in the CON group (r = −0.144, p = 0.48; Figure 3A). Furthermore, within the VA group, we discovered a significant positive correlation between the visual creativity score and the strength of FC between the left ITG and the right cuneus (r = 0.415, p = 0.003; Figure 3B), as well as the strength of FC between the left ITG and the right SOG (r = 0.621, p < 0.001; Figure 3C). However, no significant correlations were observed in the CON group between the visual creativity score and the strength of FC between the left ITG and the right cuneus (r = −0.130, p = 0.51; Figure 3B) and the strength of FC between the left ITG and the right SOG (r = −0.201, p = 0.29; Figure 3C). Moreover, we noted a significant negative correlation (r = −0.458, p < 0.001) between the weekly practice duration (measured in hours) and the strength of FC between the left ITG and the right MOG in the VA group (Figure 3D). There were no notable differences between the VA and CON groups concerning the correlation between the strength of intraregional FC and behavioral variables (training duration, daily practice hours, and psychological measurements). Additionally, in the VA group, no significant correlations were observed between the strength of interregional FC and other behavioral variables (p > 0.05).

We observed no between-group differences in terms of the correlation between intraregional FC and other psychological measurements. In the VA group, we observed no correlations between interregional FC and the duration of training, daily practice hours, or other ATTA subscales (P > 0.05).

The IPL (particularly the AG) has been proposed as a major “store house” of artistic creativity (Chakravarty, 2012). Creative cognition and creative output as conceived in the prefrontal cortex (ventromedial and dorsolateral prefrontal cortex, respectively) are relayed to the IPL and SPL reciprocally with the perceptive visual system (dorsal and ventral pathways) for both artistic production and appreciation. The ventromedial prefrontal cortex is involved in aesthetic evaluation and appreciation (Chakravarty, 2012). The VA group presented superior intraregional functional integration in terms of higher ReHo value (Wang et al., 2011) in the neural substrates of both the dorsal pathway (right SPL, right precuneus, left AG) and the ventral pathway (left ITG/FG) (Jiang and Zuo, 2016; Table 2 and Figure 1). The SPL and AG of the dorsal pathway (involved in artistic spatiality) (Seydell-Greenwald et al., 2017) serves spatial awareness, attention, spatial/motion analysis, goal-directed action, and visuomotor control (Husain and Nachev, 2007; Shomstein, 2012; Seghier, 2013). The precuneus and SPL also participate in visual imagery and other visuospatial processing (Cavanna and Trimble, 2006; Walker et al., 2011; Tres and Brucki, 2014).

As a key region connecting the occipital and parietal cortices, the ITG plays critical roles in the perception of objects, faces, and scenes (Conway, 2018) as well as in visual creativity (e.g., visual capacity, formation of higher-level complex visual representation, divergent thinking, long-term memory storage, and visual imagery) (Miyashita, 1993; Ishai et al., 2000; Mechelli et al., 2004; Sugase-Miyamoto et al., 2011; Zhang et al., 2011). The pattern of higher intrinsic FC of the left ITG-right cuneus, -right MOG, and -bilateral SOG (Table 3 and Figure 2) in the VA group is substantiated by the anatomical and structural connectivity studies of the visual system (Zhang et al., 2013; Takemura et al., 2017; Lin et al., 2020). We considered that the constellations of local and long-range connectivity changes implicate functional synergy between brain regions and can be better appreciated in the context of the neurological underpinning of visual artistic production and appreciation (Smith et al., 2003; Vartanian and Goel, 2004; Kozbelt and Seeley, 2007; Kravitz et al., 2013; Vessel et al., 2019).

VA group was of higher creativity as compared to the control group could be evidenced by their higher ATTA scores. In the VA group, we observed a significantly positive correlation between the ATTA CI score and the strength of the left ITG-right SOG FC (Figure 3A). When we consider that the ATTA CI score is a sum of four creative capacities (fluency, originality, elaboration, and flexibility), these findings indicate the consolidation of the ventral pathway in VAs facilitates the integration of various abilities in the creative process. These findings echo those in a previous study in which it was reported that network attributes in the occipital regions are predictive of individual differences in creative ability (Jiao et al., 2017).

The positive correlations between FC strength in the left ITG (with respective right cuneus and right SOG) and visual creativity score in the VA group (Figures 3B, C) partly indicate the neural underpinnings of creativity in VAs. Visual creativity refers to the appreciation and ability to produce novel esthetically pleasing visual forms (e.g., sketches, paintings, and graphic design). The process of creating these forms depends heavily on visual imagery (Heilman et al., 2003; Sack et al., 2008; Pelowski et al., 2017) and the ability to combine disparate visual representations to form new entities. The production and appreciation of visual products are subserved by the two visual pathways (Chakravarty, 2012). These results confirm that the creation of artworks depends on the visual system and particularly the ITG of the ventral pathway, which controls visual imagery, visual perception (Ishai et al., 2000), and visual attention for object recognition (Zhang et al., 2011). Object visualization ability (the ability to construct visual appearances of objects in terms of their shapes, color, and texture) in the ventral pathway and spatial visualization ability (the ability to spatial awareness and direction of movements) in the dorsal pathway both contribute to artistic creativity (Kozhevnikov et al., 2010, 2013). Scores on ATTA emphasize object visualization ability over spatial visualization ability to enrich the creative content. Our data indicates that the ventral pathway is particularly consolidated in VAs, due to strengthening of the FC between the ITG and the cuneus/SOG following long-term practice in bringing artistic notions to fruition.

Extensive training can lead to “effortless doing,” which manifests neurologically as diminished brain activity, neural oscillation, and functional connectivity (Sampaio-Baptista et al., 2015; Ji et al., 2017). One previous study of elite athletes discovered that lower amplitude signals (low- and high-frequency alpha event related desynchronization) in the ventral pathway (in the occipital and temporal areas) are an indication of processing that is more efficient than that observed in normal individuals (Babiloni et al., 2009). High-efficiency processing is characterized by a bidirectional reduction of activation in areas associated with task execution and the deactivation of regions associated with the processing of irrelevant information (Qiu et al., 2019). It has been suggested that the ITG is involved in visual object recognition and visual perception, whereas the MOG is primarily involved in the characterization of objects in terms of shape and category (Proklova et al., 2016). The occipital and temporal regions together co-constitute the ventral visual pathway (“what” route) for object recognition (Kravitz et al., 2013). The negative correlation between FC strength in the left ITG-right MOG and the duration of weekly practice (as an indicator of practice intensity) (Figure 3D) connotes “effortless doing” as a manifestation of proficiency in the execution of visual skills following the sustained practice of skills involving visual manipulation (Sheth and Young, 2016; Conway, 2018). Long-term training and the resulting proficiency in the low-level processes typically engaged in the visual arts provides a solid foundation for the creative production of artworks, which requires efficient cognitive engagement.

The current study has several limitations, which should be considered in the interpretation of our findings. First, only the ATTA was used as an indicator of general creativity. Future work should include other more specific creativity tasks (e.g., spatial visualization ability and creativity) to further explore the psychological manifestations of long-term training in the visual arts. Second, ReHo and ReHo seed-based FC can be used together to detect the synchronization of brain activity (local and long-distance) (Yan et al., 2013). The combination of these two approaches can help to identify cardinal hubs and facilitate seed selection for FC analysis (Zang et al., 2004). In the current study, we focused only on the visual system, as it is the most fundamental neural system for VAs. In the future, researchers could conduct a more thorough analysis of the FC and the neurodynamics of otherwise identified neural substrates and neural networks potentially involved in visual creativity (e.g., default mode network) (Beaty et al., 2014). These networks could be employed as heuristics by which to elucidate changes in the brain traits of VAs (Takeuchi et al., 2011).