The assessment of creativity developed by psychologists includes the characteristics of originality, fluency, flexibility, openness to new experiences, and the ability of divergent thinking (DT) (Guilford, 1950; Drevdahl, 1956). DT refers to a thought process generating various and often unconventional ideas/solutions against a single problem, whereas convergent thinking leads to conventional solutions. DT is frequently examined in neuroscientific studies of creativity, and DT tests (Guilford, 1950; Torrance, 1968) have been recognized as appropriate indicators of creative ability (see Kim, 2008; Runco and Acar, 2012).

Creativity is usually categorized into two types, i.e., domain-specific creativity, which corresponds to high quality works and achievements in a certain domain, such as visual art, music, poetry and science, and domain-general creativity, which refers to a more basic layer of creative thoughts (Sternberg, 2009). Psychometric tests correlating with domain-general creativity include Torrance Tests of Creative Thinking (TTCT) (Torrance, 1968); Alternative Uses Task (AUT) (Martindale and Mines, 1975); and Chain free association tasks (Benedek et al., 2012). To measure domain-specific creativity, various questionnaires on specific creative achievement are administered, or the subjects are given creative tasks from certain domains. The process of creative activity can be divided into generation and evaluation phases (Ellamil et al., 2012; Liu et al., 2015).

Neuroimaging experiments to investigate neural correlates of creativity have employed various paradigms and tasks, e.g., to study either domain-specific or domain-general creativity and the different phases of the creative process (generation or evaluation). Since a broader overview of neuroimaging methods including EEG has been provided elsewhere (Dietrich and Kanso, 2010; Sawyer, 2011), we focus here on functional MRI (fMRI) as it is the most widely used brain imaging technique. The studies referenced in the text below are summarized in Table 1, representative studies are also visualized in Figure 1 depicting a proposed model of neural correlates of different phases of creative activity process.

Experts in specific domains have been often considered as optimal subjects to study the effects of practice/training in creative activities on the brain. The following are examples of studies in which the brain activity of experts was compared with that of novices in task-related fMRI:

1. Planning artwork imagined from a given word was associated with stronger FC between the DMN and EN nodes than visual imagery of the alphabet or rest. The difference was greater in professional artists compared to non-artists. Visual artists may tend to simultaneously activate two processes, i.e., finding and evaluating ideas (De Pisapia et al., 2016).

2. Subjects trained in design and novice subjects were compared during task-fMRI, in which they attempted to design a new pen. In the expert group, creativity was correlated with the right-hemispheric dominance in the PFC, whereas in novices, creativity was negatively correlated with activation in the bilateral inferior parietal lobule (IPL) (Kowatari et al., 2009).

3. Total hours of musical improvisation experience were associated with increased task-specific FC of the bilateral DLPFC, dorsal premotor cortices, and pre-supplementary motor areas, whereas negative association was found with brain activity in EN during performance of improvising task compared with rest, suggesting lower demands on executive control during skilled improvisation (Pinho et al., 2014).

4. No significant difference between artists (writers and a filmmaker) and scientists (neuroscientists and molecular biologists) during a word association task was observed. Since both groups activated DMN and the same areas involved in higher-order socioaffective processing, brain cortices of very gifted artists and scientists may respond similarly (Andreasen, 2012).

5. Generation of new poetry was associated with greater deactivation of the DLPFC and intraparietal sulcus (both part of EN) in experts compared to novices, while the mPFC (DMN) was active during both phases in the both groups. Therefore, experts may prefer being in a defocused state, during which cognitive control is suspended, in order to produce outcomes more efficiently (Liu et al., 2015).

In summary, these studies consistently illustrate the pivotal role of DMN and EN. Creative performance in experts elicited greater activation and FC within DMN, whereas regions of the EN were suppressed/deactivated compared to novices. In contrast, during the evaluation phase, FC between these networks increases in both groups. Thus, long-term training in creative performance may potentially alter the connectivity of these regions.

The central role of cross-modal areas (DMN and SN) has also been demonstrated in studies searching for consistent features in brain structure associated with training in visual art. Increased gray matter density in the right precuneus was correlated with the intensity of long-term training in art school (Chamberlain et al., 2014), whereas gray matter volume of the left anterior cingulate cortex (ACC) and the medial frontal gyrus was correlated with figural-TTCT scores in university students (Chen et al., 2014). Regarding rsFC, Lotze et al. (2014) compared rsFC of creative writers with novices and observed decreased rsFC between IFG of both hemispheres as well as increased rsFC between right caudate and intraparietal sulcus in experts. On the other hand, Chen et al. (2019) compared rsFC of experts in Chinese calligraphy with novices using graph theoretical methods. They observed that the experts showed shorter characteristic path lengths and higher local efficiency in frontal and parietal cortices, limbic system, basal ganglia, and thalamus, concurrently the network measures in their regions were associated with their performance of calligraphy both in copying and creating works.

Cross-sectional studies, however, may not fully differentiate between inherent creativity traits and the effects of training. In the following two sections, we therefore focus on longitudinal studies on cognitive training demonstrating training effects on creative performance and the corresponding brain changes.

As many examples of creativity training are domain-specific, changes due to increased creative capacity have to be considered with respect to the effects of general artistic training.

Schlegel et al. (2015) observed that a three-month course in introductory drawing/painting resulted in increased figural-TTCT scores in students. DTI analysis showed reorganization of bilateral prefrontal white matter, in which fractional anisotropy (FA) decreased in art students relative to the controls who studied introductory chemistry. Furthermore, the fine-scale multivariate patterns of BOLD activity during this task, especially in the anterior cerebellum functionally related to the right upper limb cortical regions, became increasingly separable between art and non-art students, but no differences using univariate analysis were observed.

On the other hand, general music training, such as weekly keyboard training in children, can lead to improvement in motor ability (finger dexterity) associated with brain deformation changes in the primary motor area, corpus callosum and primary auditory area as measured by structural MRI over 15 months (Hyde et al., 2009). Similarly, comparison of children undergoing 2-year music training with children either involved in sports or not participating in any after-school training revealed a different rate of cortical thickening between the right and left posterior superior temporal gyrus, as well as higher FA in crossing callosal pathways connecting superior frontal, sensory, and motor segments (Habibi et al., 2017). Besides music-specific brain regions, music training may affect the function of other cognitive regions, including the EN. In a longitudinal observational study with a cross-sectional fMRI assessment by Sachs et al. (2017), children with 2-year music training showed greater activation within the EN compared to the control group, while there were no differences in performance on behavioral measures of executive function. On the other hand, in a preliminary study by Park et al. (2015), children after 15 weeks of music training showed improved executive function correlated with increased cortical thickness in the left postcentral gyrus and superior parietal lobule.

Moving to musical creativity training, recently Arkhipova et al. (2021) investigated changes in brain activation during listening to music and non-music sounds after 2-day creativity workshop in which participants composed music using non-music sounds, such as industrial noise and nature sounds. Activation in the trained group increased in DMN and decreased in EN, motor and auditory networks regardless of the sound category, while subjects’ favorability ratings toward non-musical sound samples increased, suggesting that the training broadened aesthetic perception of non-musical sounds. The tendency of change in DMN and EN is similar to aforementioned studies in high domain-general creativity (Beaty et al., 2015; Liu et al., 2015; Marron et al., 2018).

The longitudinal studies in artistic education (Park et al., 2015; Schlegel et al., 2015) therefore suggest that the emergence of creativity and/or concomitant cognitive and artistic skills may be supported by plasticity in neural functions that mediate (possibly domain-specific) sensorimotor integration, whilst more specifically creativity-oriented training, albeit a short one, may lead to changes in the functional brain networks resembling those engaged during the development of domain-general creativity (Arkhipova et al., 2021).

Another way to capture dynamic changes of brain networks is to investigate functional connectivity within and between functional networks. Several studies (Cousijn et al., 2014; Wei et al., 2014; Fink et al., 2018) have reported modulation of FC after short-term domain-general creativity training. Wei et al. (2014) demonstrated that creativity training induced an increase in rsFC between the mPFC and middle temporal gyrus (MTG) as well as in the behavioral measurement of originality. The effect was larger in individuals who originally had lower creativity. Also, Cousijn et al. (2014) investigated rsFC in adolescents after eight sessions of AUT training over 2 weeks, compared with an active control group. Although they did not find any significant difference between groups, post hoc analyses showed that change in DT performance over time course was predicted by FC between the left supramarginal gyrus and right occipital cortex, while stronger FC between the MTG and bilateral postcentral gyrus was associated with better DT performance across both groups. In another longitudinal fMRI study by Fink et al. (2018), successful 3-week verbal DT training measured by AUT was reflected by decreased rsFC within DMN, sensorimotor and auditory networks, as well as in the attention network. Taken together with the related study of task-fMRI after the same training (Fink et al., 2015), in which activation changes were observed in the left IPL and the left MTG, the authors concluded that these global changes in rsFC may support the activity in task-specific brain regions, thus suggesting that the analysis of rsFC can be useful tool for indexing effects of DT training.

As shown above, it can be said that certain educational programs at/after school and short-term training can stimulate creativity and plastic changes in the associated regions in the brain, analogous to plasticity induced by physical exercise or working memory training.

Besides specific creativity training, recently Zulkifli et al. (2025) suggest that parenting style affects offspring’s creativity measured by DT as well as brain development in the right supramarginal gyrus, which is a node in the SN and DMN.

Other approaches to enhance creativity include EEG-neurofeedback (Raymond et al., 2005; Gruzelier et al., 2014a, b; Imperatori et al., 2017) and mindfulness meditation (Zabelina et al., 2011; Colzato et al., 2012; Ding et al., 2014) but there are few studies (simultaneously) evaluating the brain changes associated with such interventions [for an overview of effectiveness of creativity enhancement methods in adults, see meta-analysis by Haase et al. (2025)]. For example, an interventional study by Colzato et al. (2012) suggested that mindfulness meditation training improves DT. Meditation is known as a method to reduce stress and promote wellbeing (Zabelina et al., 2011; Colzato et al., 2012; Ding et al., 2014; Brewer et al., 2011) as well as physical and mental relaxation (Deepak, 2019). An fMRI study by Brewer et al. (2011) comparing trained meditators and novices during in-scanner meditation showed that nodes of the DMN (mPFC and PCC) were relatively deactivated in experts. The authors also reported that experience in meditation was associated with decreases in activity in the brain regions comprising DMN during rest, as well as stronger rsFC between PCC and dorsal ACC, and between PCC and DLPFC. The rsFC associated with mediation thus resembles the aforementioned rsFC pattern related to high-creativity prediction model by Beaty et al. (2018).

Taken together with the studies of neural correlates of artistic performance and high creativity mentioned in the first part of this review, it can be suggested that a relaxed state of mind or suppressed self-referential thinking (as reflected by lower activity of DMN during resting state, Brewer et al., 2011), are important factors for one’s best creative performance and decision making. At the same time, effective creativity training seems to promote such patterns in brain activation.