Brain Basis of Creative Conceptual Expansion

The primary aim of this fMRI study was to replicate the findings of a previous study investigating the brain basis of creative conceptual expansion. The prior research had developed a novel event-related experimental design to examine this cognitive process. The original study identified specific brain regions associated with creative conceptual expansion, namely the anterior inferior frontal gyrus, the temporal poles, and the lateral frontal pole. These regions are known to be involved in the access, storage, and relational integration of conceptual knowledge. The current study aimed to verify if these findings could be reliably replicated, a crucial step given the ongoing replication crisis in psychology and neuroscience. The ability to successfully replicate these findings would significantly strengthen the understanding of the neural mechanisms underpinning creative thought processes and the reliability of the previously employed experimental methods. The replication of these results would provide a robust foundation for future research into the neural underpinnings of creativity.

The second objective was to investigate the neural correlates of individual differences in creative conceptual expansion. This involved comparing brain activity patterns between participants identified as high-creative and low-creative. The hypothesis was that high-creative individuals would exhibit greater activity in brain regions associated with semantic cognition and the salience network during tasks involving creative conceptual expansion. This part of the study aimed to identify the specific neural mechanisms underlying individual variations in creative abilities. The exploration of individual differences would provide valuable insights into the factors contributing to variations in creative potential. The results would help in developing more nuanced models of creative cognition, moving beyond general observations to account for individual variations in brain function and cognitive styles. The research also planned to examine this in light of classical information processing theories such as flat associative hierarchies, defocused attention, and cognitive disinhibition; these theories posit different information processing biases that could explain why some people are more creative than others.

The study's hypotheses were informed by existing literature on the neuroscience of creativity. Contemporary global approaches emphasize the roles of large-scale brain networks, particularly the default mode network (DMN) and the central executive network (CEN), in creative thinking. The DMN is linked to spontaneous cognition, while the CEN is associated with goal-directed processes. Some theories suggest that creativity involves the interplay between these networks, with the DMN generating ideas and the CEN selecting among them. However, the study acknowledged the relative scarcity of contemporary neuroscientific models specifically addressing the brain basis of individual differences in creativity. To address this gap, the researchers considered three classical information processing bias theories: flat associative hierarchies (Mednick, 1962), defocused attention (Mendelsohn, 1974), and cognitive disinhibition (Martindale, 2007). These theories offered potential explanations for individual variations in creative performance and informed the study's design and interpretations of the fMRI data. The study aimed to investigate the potential neural substrates of these biases by examining different brain activation patterns between high- and low-creative individuals.

The study utilized a sample of 110 female university students from Justus Liebig University Giessen, Germany. This homogenous sample was chosen for this study. This participant selection method might limit the generalizability of the findings to other populations. Participants undertook two key tasks to measure their creativity: the Alternate Uses Task (AUT) and the Vocabulary Scale of the Hamburg Wechsler Intelligence Test for Adults (HAWIE). The AUT involved generating multiple uses for common objects (brick, newspaper, bottle, tire, umbrella) within a time limit of 2 minutes per trial. The HAWIE vocabulary scale assessed intellectual ability by requiring participants to define words of increasing difficulty. These scores were converted to standardized values considering participant age. A subsample of 34 participants from this group later underwent fMRI scanning. Pre-scanning, participants from this subset completed a 10-minute practice session to familiarize themselves with the tasks. Prior to fMRI scanning, the participants' creative abilities were also assessed by an Abbreviated Torrance Test for Adults (ATTA). The participants were then divided into high- and low-creative groups based on their performance on these creativity assessments. Importantly, there were no significant differences in age or WST IQ scores between these two groups, suggesting that the observed differences in brain activity were not confounded by age or general intelligence. Ethical considerations were addressed, with all participants providing informed consent and receiving compensation for participation. The study's procedures were approved by the Ethics Commission of the German Psychological Society (DGPs).

The study employed a 2 x 2 repeated measures factorial design, replicating the methodology of a previous fMRI study (Abraham, Pieritz, et al., 2012). The factors were Task Type (divergent vs. control) and Cognitive Demand (high vs. low). Divergent tasks included the Alternate Uses Task (AUT, high-demand – DH; low-demand – DL) and the Object Location Task (OLT). The control tasks were 2-back (high-demand – CH) and 1-back (low-demand – CL) working memory tasks. Each fMRI run comprised 20 trials per condition and 8 resting baseline trials, pseudo-randomly interspersed. Visual stimuli were presented in a pseudo-randomized order to counterbalance trial transitions. The total fMRI experimental session lasted about 38 minutes (25 seconds per trial). Following the fMRI session, a feedback session was conducted where participants reported the responses they generated during the scanning session. This allowed for verification of response accuracy against the experimenter's independent record. This process of recording responses in the scanner and then verifying them during a post-scan feedback session aimed to minimize response errors and ensure accurate data collection. The use of both high and low demand tasks within each condition allowed for a more detailed analysis of the cognitive processes involved in creative conceptual expansion, while also controlling for differences in task difficulty.

Functional and anatomical magnetic resonance imaging (MRI) was conducted using a 1.5 Tesla whole-body tomography system (Siemens Symphony) at the Bender Institute of Neuroimaging (BION). Participants were positioned supine in the scanner, with their right index finger on a response box. Careful stabilization and cushions minimized movement. Earplugs attenuated scanner noise. Structural image acquisition involved 160 T1-weighted sagittal images (MPRAGE, 1 mm slice thickness). Functional imaging involved one run with 902 volumes recorded using a T2*-weighted gradient-echo-planar imaging sequence (EPI). The data preprocessing mirrored the original study (Abraham, Pieritz, et al., 2012), utilizing the LIPSIA software package. This included motion correction, sinc-interpolation to correct for slice timing differences, application of a temporal highpass filter to remove low-frequency signal changes, spatial smoothing with an 8 mm FWHM Gaussian filter, and normalization to Talairach standard space with a final voxel size of 3x3x3 mm. The consistency in data preprocessing ensured that the results could be directly compared to the results of the earlier study, and the use of established software minimized the likelihood of introducing biases during data analysis. The carefully controlled methods minimized noise and maximized the accuracy of data analysis.

This fMRI study successfully replicated previous findings on brain activation patterns during creative conceptual expansion. The study used the same experimental design as the original study by Abraham, Pieritz, et al. (2012) which employed a 2 x 2 repeated measures factorial design examining task type (divergent vs. control) and cognitive demand (high vs. low). The current study confirmed the involvement of key brain regions previously identified as being crucial for creative conceptual expansion. These included the anterior aspects of the lateral inferior frontal gyrus (BA 45 and 47), the temporal pole (BA 38), and the lateral anterior prefrontal cortex (BA 10). These regions are implicated in the access, storage, and relational integration of conceptual knowledge. The remarkable similarity in the activation patterns between the current study and the original study strongly supports the robustness and reliability of the findings. This replication is significant given the current emphasis on reproducibility in scientific research. The consistent findings across both studies provide strong evidence for the specific neural mechanisms underlying creative conceptual expansion and highlight the validity of using the established methodology and paradigms for investigating brain activity associated with this complex cognitive process.

Beyond replicating previous findings, the current study revealed additional brain regions significantly activated during creative conceptual expansion. These previously unreported activations, identified through a contrast analysis comparing the divergent-high (DH) and control-high (CH) conditions and subsequently using an inclusive mask in the analysis of the divergent-high (DH) versus divergent-low (DL) contrast, include posterior regions of the inferior frontal gyrus (BA 44), the middle frontal gyrus (BA 8), the anterior cingulate cortex (BA 24), dorsomedial prefrontal cortex (BA 8, 9, 6), the inferior parietal lobule (BA 40), and regions of the basal ganglia (putamen, globus pallidus). The pattern of activation was more strongly lateralized to the left hemisphere, consistent with the original study. The identification of these additional brain areas expands our understanding of the complex neural networks involved in creative conceptual expansion. The unexpected activations suggest that creative cognition may involve a more widespread network of brain regions than previously thought, and further investigation is warranted to understand their precise roles in this cognitive process. The lateralization to the left hemisphere further strengthens findings implicating semantic processing in creative conceptual expansion.

The fMRI data analysis employed both frequentist and Bayesian approaches. For the replication portion of the study, a second-level random-effects analysis using one-sample t-tests examined differences across all subjects. The inclusive mask technique focused analyses on regions activated in the divergent-high versus control-high contrast (DH > CH). Multiple comparisons were corrected using double thresholding (z > 3.09, p < .001, and cluster-size and cluster-z-value thresholding at p < .01). For analyses of individual differences, Bayesian analyses were performed, comparing high- and low-creative groups using contrasts of Divergent-High (DH) > Divergent-Low (DL) and Divergent-High (DH) > Control-High (CH). Inclusive masks specific to each creativity group were used. A conjunction analysis combined the DH > DL and DH > CH contrasts to pinpoint regions specifically involved in conceptual expansion. A threshold of 99.5% probability (Bayesian) was used, along with a minimum cluster size of 10 voxels. The combination of frequentist and Bayesian approaches strengthens the reliability and generalizability of the results, enabling a more comprehensive understanding of the neural substrates involved in creative conceptual expansion. The use of rigorous statistical methods addresses the challenges of multiple comparisons and provides confidence in the reported findings.

The study investigated individual differences in brain activity during creative conceptual expansion by comparing high- and low-creative groups. Bayesian analyses were conducted on fMRI data, focusing on brain activity patterns selectively engaged during the high-demand Alternate Uses Task (DH) relative to both the low-demand Object Location Task (DL) and the control 2-back task (CH). This approach aimed to isolate brain regions specifically associated with creative conceptual expansion and to examine how this activation differed between the two groups. Surprisingly, no significant brain activity differences were found for the low-creative group relative to the high-creative group. In contrast, the high-creative group showed significantly greater activity in several key brain regions. This suggests that the neural underpinnings of creative conceptual expansion might be more strongly associated with heightened activity in specific regions rather than a suppression of activity in low-creative individuals. This pattern of results highlights the importance of considering individual differences in the neural correlates of creativity and might point to distinct neural mechanisms associated with high levels of creative ability.

The high-creative group demonstrated significantly greater activation compared to the low-creative group in specific brain regions during creative conceptual expansion. These regions included the anterior lateral inferior frontal gyrus (BA 45, 47) in both the left and right hemispheres, the temporal pole (BA 38), and the posterior middle temporal gyrus (BA 37). These regions are known to be involved in accessing and processing semantic information. The increased activity in these areas suggests that highly creative individuals might allocate more neural resources towards semantic operations during creative tasks, aligning with theories proposing that creativity involves a broader and more flexible retrieval of semantic knowledge. The involvement of the anterior lateral inferior frontal gyrus, a region known to be involved in semantic processing and working memory, aligns with theories of creativity that emphasize the importance of accessing and manipulating information from long-term memory. The lack of significant differences between the groups in the number of responses generated implies that these differences in brain activity are not simply due to differences in the quantity of ideas produced but rather in the neural mechanisms employed in generating creative ideas.

The findings also indicated a significant role for the salience network in individual differences in creative conceptual expansion. The high-creative group displayed greater activation in the insula and dorsal anterior cingulate cortex (dACC), core regions of the salience network. This suggests that the salience network plays a role in the process of creative conceptual expansion, likely by modulating the interaction between the default mode network (DMN) and the central executive network (CEN). This is interesting as it contrasts with existing global-level interpretations of creativity which primarily focus on the DMN and CEN interactions. While the study doesn't directly support or refute these broader brain network models, the finding raises intriguing questions about the specific role of the salience network in facilitating creative thinking. The results do seem to support Mednick's theory of flat associative hierarchies and Mendelsohn's idea of defocused attention, but it is not direct proof against Martindale's cognitive disinhibition theory due to the nature of the experimental design. The involvement of the salience network underscores the complexity of creative cognition, pointing to the importance of integrating multiple brain networks in understanding this process. Future studies could further explore the dynamics of these network interactions using global approaches combined with multiple measures of creative cognition to provide a more detailed and nuanced understanding of the neural correlates of creativity.

The study successfully replicated previous fMRI findings on the neural correlates of creative conceptual expansion, confirming the involvement of the anterior inferior frontal gyrus, temporal poles, and lateral frontopolar cortex. Furthermore, it extended these findings by identifying additional brain regions activated during this process, including areas within the inferior parietal lobule and basal ganglia, suggesting a more distributed neural network than previously recognized. Crucially, the study's investigation into individual differences demonstrated that high-creative individuals exhibited significantly greater activity in regions associated with semantic processing (anterior inferior frontal gyrus, temporal pole, posterior middle temporal gyrus) and the salience network (insula, dorsal anterior cingulate cortex) during creative conceptual expansion. This highlights the importance of considering individual variations in brain activity when studying creativity. These findings support existing theoretical frameworks suggesting that individual differences in creativity might stem from variations in the efficiency and flexibility of semantic processing and attentional control. The study's results provide substantial support for a local approach to understanding the neural basis of creativity, focusing on specific brain regions involved in semantic processing and network modulation. However, the lack of a direct test between the three competing theories (Mednick, Mendelsohn, and Martindale) means we can only infer a degree of support, and these findings should not be considered disproving any of the models.

The study acknowledges limitations, including the use of a homogenous sample of female university students, potentially restricting the generalizability of the findings. The absence of post-fMRI difficulty ratings, a component of the original study, also represents a methodological limitation, preventing a comprehensive evaluation of the subjective experience of cognitive demand across the different task conditions. The observed decrease in the number of responses generated in the fMRI session compared to a pre-scanning session, possibly attributed to forgetting, was another aspect to note. Despite these limitations, the robust replication of the core findings strengthens the confidence in the results. Future research should address these limitations by using a more diverse sample, including measures of subjective cognitive workload, and improving response memory protocols. Future research directions suggested include examining other relevant cognitive operations in creativity, such as creative imagery and the influence of previously activated knowledge on original response generation. These investigations could further elaborate upon the dynamics of conceptual expansion and extend our understanding of the multifaceted nature of creative cognitive processes. The study serves as a foundation for future research utilizing both local and global approaches, incorporating multiple measures of creative cognition to achieve a more thorough understanding of creative thinking’s intricacies.