There likely exist myriad cognitive skills necessary to produce something both “novel and useful.” These skills probably manifest differentially within various domains (e.g., visual art vs. scientific discovery), and common creativity might differ substantially from creative genius (Dietrich, 2004). Indeed, creative productivity has been studied across numerous activities, like musical improvisation (Bengtsson et al., 2007; Berkowitz and Ansari, 2008; Brown et al., 2006; Limb and Braun, 2008); visual art perception and esthetics (Bhattacharya and Petsche, 2002; Cela-Conde et al., 2004; Kirk et al., 2009); dance (Fink et al., 2009b); neural comparisons between groups of artists, musicians, and matched controls during creative performance (Bhattacharya and Petsche, 2005; Gibson et al., 2009); and evaluation of combination of these across modalities and subjects (Petsche et al., 1997). In laboratory settings, the assessment of subject engagement in creative tasks is made mostly by tests of divergent thinking (DT), the process by which one extrapolates many possible answers to an initial stimulus or target data set (Guilford, 1967). Other central constructs include fluid intelligence (Cat-tell, 1943), insight—the flash of recognition that a problem is solved (Jung-Beeman et al., 2004), and “flow” defined as “when the person is fully immersed in what he or she is doing by a feeling of energized focus, full involvement, and success in the process of the activity” (Csikszentmihalyi, 1996). In all likelihood, some combination of these and other cognitive processes underlies the creative process, which involves a focused attention to the exclusion of other competing stimuli (i.e., flow); divergence of ideas to numerous possible novel solutions to a given problem (i.e., divergent reasoning); if one is lucky a flash of insight, if not then convergence on the best solution (i.e. utility); and perseverance in the face of social acceptance or resistance (i.e., personality variables). Comprehensive neuroscience research is evolving to incorporate combination of these and other cognitive and personality measures to address the complex construct of creativity.

Neurological inquiries regarding creativity have tended to focus on whether the frontal lobes are engaged or whether more posterior brain regions (Heilman et al., 2003) or subcortical structures such as the basal ganglia are more predominant (Dietrich, 2004). Such efforts are based largely on data gleaned from neurological and psychiatric patients (Pollack et al., 2007). Indeed, de novo artistic expression has been associated with left frontotemporal (Finkelstein et al., 1991) and right temporal lobe epilepsy (Mendez, 2005), several case studies of frontotemporal dementia (FTD) (Miller et al., 1998, 2000; Thomas Anterion et al., 2002), a case of Parkinson's disease treated with dopaminergic agonists (Schrag and Trimble, 2001), and a single case of subarachnoid hemorrhage (Lythgoe et al., 2005). Subsequent systematic study of artistic ability associated with the various dementias found no general increase in creativity to be linked with FTD, with the authors noting that “despite the existence of these isolated patients with increased artistic production, however, apathy leading to diminished creativity is more clinically typical of patients with FTD, suggesting that these case studies may be the exception rather than the rule” (Rankin et al., 2007).

Several electroencephalography (EEG) studies provide tantalizing support that imaging of the creative experience is both possible and informative to understanding the interactions of distributed neural networks. Early EEG studies demonstrated that highly creative individuals differed from normal controls in (1) greater activity within right parietotemporal areas, (2) higher alpha activity during analogs of “inspiration,” and (3) greater tendency to present physiological over-response (Martindale and Greenough, 1973; Martindale and Hasenfus, 1978; Martindale and Hines, 1975). In the middle phase of EEG studies, researchers described greater “dimensional complexity” over central and parietal cortices in subjects engaged in DT (Molle et al., 1996). Similarly, one study that compared gifted, intelligent, creative, and average individuals found lower levels of mental activity in highly creative subjects when compared to average individuals when engaged in the solution of creative problems (Jausovec, 2000). This same group (Jausovec and Jausovec, 2000) found that EEG coherence (during “rest” with eyes open) was significantly related to creativity scores, particularly across the right hemisphere. Finally, one researcher studied healthy males and found that good creative DT performance (N = 15) related to increased centroparietal interhemispheric connectivity and greater right hemisphere interconnectivity (Razumnikova, 2000). This same researcher found different patterns of interhemispheric connectivity and amplitude in men and women, suggesting significant sex-mediated differences in creative cognition (Razumnikova, 2004). More recently, this researcher found increased power in the frontal cortex and increased desynchronization over the posterior cortex associated with performance during a verbal insight task (Razumnikova, 2007). Additional EEG work is reported for the rarely studied realm of scientific hypothesis generation. Twenty-five gifted and twenty-five age-matched controls were compared, with results suggesting increased information transfer within left posterior brain regions of the gifted children compared to controls (Jin et al., 2006).

One research group has contributed much to the newest phase of EEG studies of creativity, with initial studies showing lower levels of cortical arousal during creative problem solving, and stronger alpha synchronization in centroparietal cortices associated with more original responses (Fink and Neubauer, 2006). This same group found the creativity–alpha power relationship to be mediated by the personality characteristic of Introversion–Extraversion (Fink and Neubauer, 2008). Finally, in a combined EEG/fMRI study, they were able to interpret EEG alpha band synchronization, particularly within the frontal lobe, with active cognitive processes rather than cortical idling (Fink et al., 2009a). Thus, there is considerable heterogeneity of findings across EEG studies of creative cognition, making it difficulty to draw robust conclusions regarding the impact or direction of alpha activity, synchronization and localization of these factors within frontal, posterior, or even lateralized hemispheric cortices. Moreover, when such relationships are found, they appear to be mediated by giftedness (Jausovec, 2000; Jin et al., 2006), personality variables (Fink and Neubauer, 2008), and sex (Razumnikova, 2004).

The neurobiology of creativity also has been investigated with brain imaging techniques including regional cerebral blood flow, single photon emission computerized tomography (SPECT), positron emission tomography (PET), and functional magnetic resonance imaging (fMRI). One early study (Carlsson et al., 2000) was undertaken in 12 healthy male subjects stratified by either high or low scores on the creative functioning test (Smith and Carlsson, 1990). Blood flow measures were compared during performance of verbal fluency and DT. The highly creative group was characterized by bilateral frontal activation during DT compared to predominantly left hemisphere activation in the low creative group. Interestingly, better performance on the DT task was negatively correlated with higher activity within superior frontal regions. Such inverse correlations are suggestive of neural or network efficiency and have also been reported in neuroimaging studies of intelligence (Haier et al., 1988, 1992; Neubauer et al., 2004), although these efficiencies are now hypothesized to exist mainly for the frontal lobes (Neubauer and Fink, 2009). In another study, SPECT was used with 12 highly creative subjects while performing figural and verbal creativity tasks. These authors found a positive relationship between the creativity index and cerebral blood flow in the right postcentral gyrus, bilateral rectus gyrus, right inferior parietal lobule, and right parahippocampal gyrus (Chavez et al., 2004). PET was used to study nine healthy subjects as they performed verbal insight tasks (Starchenko et al., 2003). These authors found that the creative process activated left Brodmann area (BA) 40 and the cingulate gyrus (BA 32). This same group used PET to study normal subjects as they performed verbal creativity tasks and observed brain activations in the left parietotemporal brain regions (BAs 39 and 40) (Bechtereva et al., 2004). One fMRI study attempted to localize creative story generation within the brains of a cohort of eight normal subjects (Howard-Jones et al., 2005). When creative story generation was contrasted with uncreative story generation, significant activations were observed within bilateral medial frontal gyri (BAs 9 and 10) and the left anterior cingulate (BA 32). A novel study had subjects generate responses to the Rorschach inkblots (Asari et al., 2008). These researchers found that when unique responses were compared to more frequently generated responses, greater activations were observed within the right temporal pole (BA 38). When a less stringent threshold was used, additional regions associated with unique blot generation were identified within the left orbitofrontal region (BA 11), left cingulate (BA 32), and the left parietal cortices (BA 39). Thus, across functional studies, there appears to be some convergence, as noted previously, suggesting importance of the parietal cortex (BAs 39 and 40) to the creative process, the cingulate (BA 32) involved with internal selection, and frontal regions being engaged relevant to task complexity (BAs 8, 9, and 47) (Starchenko et al., 2003).

Undoubtedly, localization of creativity to certain regions of the brain is hampered by the lack of a systematic framework by which to empirically approach such a complex construct. For example, a wealth of research exists from the psychometric literature linking creativity to intelligence and personality variables (Batey and Furnham, 2006), yet these measures are rarely included in modern neuroimaging studies to assess the discriminant validity of the creativity measure of interest (e.g., DT). Most studies conflate creativity with a discrete cognitive process as opposed to assuming that a given cognitive process (e.g., DT, insight, fluid intelligence) is but one of many components making up the creative act. What these studies highlight is that a construct as complex as creativity will never be “localized” in the brain—be it the right hemisphere, anterior cingulate cortex, or other locus. Rather, individual findings will be dependent upon the task used as a “proxy” measure (e.g., insight, DT, convergent thinking), the population under scrutiny (e.g., college undergraduates, experts), and even methodological issues related to structural versus functional brain characteristics.

Neuroscience inquiries of creativity show a muddled picture likely related to subject, modality, and metric issues: Lesion studies tend to localize creativity to the anterior frontal and temporal poles; EEG studies show both “higher” and “lower” activation and more diffuse or focal activity based on task and subject characteristics; functional imaging studies show a tendency toward frontal, parietal, and cingulate localization. All studies have relatively small samples limiting statistical power. The current report attempts to address some of these shortcomings by (1) administration of psychometrically valid measures of intelligence, personality, and creativity (i.e., DT, creative achievement) to a large cohort of healthy subjects, (2) undertaking the first structural magnetic resonance imaging study linking constructs central to creativity to cortical thickness, and (3) linking our results to previous studies across the neuroscientific, behavioral neurology, and psychometric literatures.

We found several discrete clusters at P < 0.01 that had a negative correlation with CCI and cortical thickness, which indicated decreased cortical thickness in relation to higher CCI scores (Fig. 1a–c and Table I). In the left hemisphere, the lingual gyrus (P = 0.0192) was negatively correlated with CCI. Several regions in the right hemisphere were also negatively correlated with CCI, including the following: the fusiform (P = 0.0008), the cuneus (P = 0.0006), angular gyrus (P = 0.0011), and a cluster with the largest surface area that was composed of vertices from three regions, the inferior parietal, lateral occipital, and superior parietal (P = 0.0001), with the centroid in the inferior parietal region. One region, the right posterior cingulate, correlated positively with the CCI (P = 0.0012), indicating increased cortical thickness related to higher CCI scores.

For the CAQ, there was one area for which lower cortical thickness was related to higher scores on the CAQ: the left lateral orbitofrontal region (P = 0.0065). Similarly, there was one area where higher cortical thickness was related to higher scores on the CAQ in the right angular gyrus (P = 0.0007) (Fig. 2a–b and Table II).

The inverse relationships between cortical thickness and the creativity measures in the present study speak to the possible importance of efficient information flow among brain areas. These results can be seen as consistent with our previous study showing inverse relationships between the neurometabolite NAA and creativity in a subset of the same cohort as presented here (Jung et al., 2009), and with a previous study showing decreased cortical thickness associated with functional activation in a cohort of older children (Lu et al., in press). Similarly, developmental studies of intelligence have shown accelerated cortical thinning in children (age 7–19) with the highest levels of intelligence in regions including the dorsal and rostral frontal lobes (Shaw et al., 2006). In studies spanning normal development, the parietal lobes have been found to thin most during adolescence (Sowell et al., 1999a), with the frontal lobes thinning most during late adolescence and early adulthood (Sowell et al., 1999b). The mechanism for such thinning has been postulated to involve more focused functional activation during skill acquisition as plasticity decreases and efficiency increases (Durston and Casey, 2006; Durston et al., 2006). These results, along with the current findings, would suggest that development of cognitive capacity (including creative capacity) can be associated with lower levels of cortical thickness in discrete regions of the brain, especially within younger age cohorts of which the present sample (lower age range = 18) at least partially overlaps. In applying this viewpoint to our cortical thickness results, a possible interpretation of our findings is that the generation of novel, original ideas is associated with less cortical thickness within frontal and (certain) posterior cortical regions, requiring higher functional activation to initiate cognitive control.

The distribution of significant areas throughout the brain, found in the current study, suggests that information flow among brain areas may be a key to creativity. We proposed a similar concept for an intelligence network based on a wide range of structural and functional neuroimaging studies (Jung and Haier, 2007), and some overlap in regions bears discussion. For example, cortical thickness within regions of the right angular gyrus (BA 39) was found to predict performance on both the CCI (negative) and CAQ (positive). In studies of intelligence, a wide range of research found correlations with right BA 39 across structural imaging modalities (Shaw et al., 2006), PET (Esposito et al., 1999), and fMRI (Atherton et al., 2003; Lee et al., 2006). Indeed, other researchers (Jung-Beeman, 2005) have hypothesized the angular gyrus to be important to semantic activation, a process by which “first-order” associations are made. Interestingly, the right semantic field is specifically hypothesized to maintain weak semantic links, “including distant and unusual semantic features, features that seem irrelevant to the context, and secondary word meanings … rife with ambiguity” (p. 514). Along with other studies showing posterior parietal EEG activation associated with original ideational generation (Fink et al., 2009a; Fink and Neubauer, 2006; Grabner et al., 2007) and insight (Jung-Beeman et al., 2004), this parietal focus points to bottom–up activation of attentional and analogical knowledge stores (semantic/visuospatial) necessary for creative problem solving.

There were two regions where we saw increased cortical thickness related to creative capacity: the first in the right cingulate cortex (BA 24), related to performance on DT measures; the second within the right angular gyrus (BA 39), related to creative achievement (CAQ). Previous creativity studies report relationships between various creativity measures and activations within the cingulate gyrus and parietal cortex (Chavez et al., 2004; Howard-Jones et al., 2005; Molle et al., 1996; Razumnikova, 2000). Specialized cognitive activities ascribed to the cingulate cortex include monitoring unfavorable performance outcomes, response errors, and response conflicts (Ridderinkhof et al., 2004). Most salient to the current discussion, one study found greater activation in the left inferior parietal and angular gyri when subjects performed verbal alternate uses tasks (Fink et al., 2009a). The authors interpret this finding as consistent with the inferior parietal cortex being associated with verbal working memory processing, in particular the phonological store (Baddeley, 2003). As our results were found for creative achievement (CAQ), they were not task related per se, although perhaps reflecting homologous working memory processes associated with the visuospatial sketchpad.

NAA is often considered to be a proxy measure for neuronal mass (Barker, 2001). Thus, the finding of increased cortical thickness correlates to DT in the right cingulate (BA 24) might appear to contradict our previous research showing inverse NAA-DT relationships in a largely overlapping cohort (Jung et al., 2009). However, NAA, while found predominantly in neurons, serves numerous roles both within neurons and surrounding oligodendrocytes including the following: cellular osmolyte, storage vehicle for aspartate and glutamate, metabolic precursor of the excitatory dipeptide N-acetylaspartyl-glutamate, involvement in neuronal– glial signaling, participation in myelin formation, and molecular water pump (Baslow, 2003). Moreover, cortical thickness measures represent a broad range of tissue types—not strictly neuronal. Indeed, it is plausible that increased glial cell content would result in higher cortical thickness and lower levels of NAA within the same tissue. It is of note that Albert Einstein, a creative icon of the 20th century, had parietal lobes, which were 15% wider than control subjects (Witelson et al., 1999), while left BA 39 within the parietal cortex was found to have significantly higher glial:neuronal ratio on histology (Diamond et al., 1985). Future research focused on myoinositol, a glial marker within the proton spectrum, would potentially help disentangle this complex relationship between brain biochemistry, cortical thickness, and creativity.

Several inverse relationships were observed between cortical thickness measures and creativity as measured with the CCI. For the left hemisphere on the CCI, this was limited to the lingual gyrus (BA 18). Interestingly, a recent study reports that the lingual gyrus was preferentially activated to process novel events that were outside the focus of spatial attention (Stoppel et al., in press). These authors interpret this finding to support the lingual gyrus acting as a “novelty detector at early perceptual level” (p. 10). As two of the three DT measures that comprised the CCI were timed novel drawing tasks, it is plausible that structural differences within the lingual gyrus accounted for some of the behavioral differences on the CCI. Other regions in which inverse relationships were found for the CCI, in the right hemisphere, included the cuneus (BA 18), fusiform (BA 19), inferior parietal (BA 19), and angular gyrus (BA 39). Taken together, it is difficult to link the occipital and parietal regions to previous research in creativity, as no studies have focused on activation patterns clustered so posterior in the brain, although several EEG studies have found a preponderance of centroparietal involvement in high creative performance (Fink and Neubauer, 2006; Molle et al., 1996; Razumnikova, 2000). Similarly, the relationship between CCI and BA 39 is consistent with previous research showing relationships between cortical activation and performance on the Torrance Test of Creative Thinking (Chavez et al., 2004). However, no previous research has specifically assessed creative achievement in young adults, and the results should be interpreted with caution until replicated.

Finally, BA 11, within the left orbitofrontal cortex, was inversely related to the CAQ. This finding is of note given previous research showing creative increases in some patients suffering from FTD—a disease preferentially affecting the orbitofrontal cortex and temporal poles (Miller et al., 1998, 2000; Miller and Hou, 2004), although this relationship does not hold across a case-controlled cohort of patients with FTD (Rankin et al., 2007). The left orbitofrontal cortex is also of interest as it was one of the unconstrained regions found to predict unique responses when subjects responded to Rorschach inkblots in an fMRI paradigm (Asari et al., 2008). Indeed, both visual and auditory inputs project to the orbitofrontal cortex via the superior temporal sulcus and temporal pole. In turn, the orbitofrontal cortex projects back to regions distributed broadly throughout the cerebral cortex, including the amygdala (Barbas, 2007), anterior cingulate cortex (Carmichael and Price, 1996; Price, 2006), basal ganglia, and other cortical regions (including the parietal cortex) (Rolls, 2008; Rolls and Deco, 2002). This network provides routes by which behavior (Rolls, 2005) and memory (Rolls and Xiang, 2005) are influenced. In both human and nonhuman primates, functioning of the orbitofrontal cortex appears to be critically involved in distinguishing rewarding from unrewarding stimuli as well as in altering behavior when reinforcement contingencies change (Rolls and Grabenhorst, 2008). This role of the orbitofrontal cortex is seen to play a critical evolutionary role by allowing maximal flexibility in possible actions. Indeed, by leveraging instrumental reinforcers in specifying goals for actions (and subsequent emotions), the orbitofrontal cortex guides the efficient matching of behavior to environmental contingencies (Rolls, 2005). Interestingly, a region of the primate orbitofrontal cortex responds to novel but not familiar visual stimuli (Rolls et al., 2005), a finding relevant to discussions of creativity.