Neural, genetic, and cognitive signatures of creativity

Creativity, a crucial element of individual problem-solving and societal advancement, is often measured through divergent thinking (DT)—the ability to generate multiple ideas for open-ended tasks. While previous neuroimaging studies have identified brain regions associated with DT, inconsistencies exist, potentially due to the multifaceted nature of creativity, demanding the coordination of multiple cognitive processes. This complexity has hindered the development of a unified framework integrating neural, genetic, and cognitive aspects. This study aimed to address this gap by employing MVPA to identify a neural pattern predictive of DT across two independent samples, and to validate this pattern through cognitive decoding, genetic data, and large-scale resting-state fMRI. The researchers hypothesized that a comprehensive neural signature of DT would reveal the complex interplay of brain networks involved and its underlying genetic and neurobiological mechanisms. The study's findings could significantly advance our understanding of creativity's neurobiological foundations.

Existing research on the neural correlates of divergent thinking (DT) using functional magnetic resonance imaging (fMRI) has yielded mixed results. While studies consistently report activation in the prefrontal cortex regions (DLPFC, VLPFC, ACC), activation in other brain areas such as the right angular gyrus, left fusiform gyrus, and left middle temporal gyrus has also been reported, highlighting the complex network involvement. These inconsistencies are likely due to the high-order cognitive nature of creativity, which relies on several processes such as free association, working memory, and inhibitory control. The dual-process theory of creativity suggests that DMN (supporting idea generation) and FPCN (supporting evaluation and selection) work collaboratively. This study leverages cognitive decoding techniques to connect activation patterns with specific psychological processes, further examining the role of genetics and neurotransmitters in DT's neurobiological underpinnings. The integration of these perspectives into a unified framework remains a crucial area requiring further research.

The study employed task-based fMRI data from two independent samples (Sample 1: n=55; Sample 2: n=30). Participants performed a modified version of the Alternative Uses Task (AUT), involving both novel use (NU, DT condition) and general use (GU, control condition) prompts for various objects. Linear support vector machines (SVMs) were used for MVPA to classify NU vs. GU conditions. A 10x10 fold cross-validation was performed to assess model accuracy, sensitivity, specificity, and AUC. The generalization performance of the models was validated using each sample as a validation set for the other. Bootstrap tests were used to identify significant features. The resulting DT brain pattern was then analyzed using cognitive decoding (correlating with Neurosynth meta-analytic maps) to understand the associated cognitive functions. Further neurobiological analyses involved correlating the DT brain pattern with gene expression data from the Allen Human Brain Atlas and neurotransmitter receptor/transporter maps from PET studies. Relevance vector regression (RVR) was used to predict individual DT scores from task-based fMRI data and resting-state functional connectivity data from three large independent samples (SLIM, GBB, BBP). Permutation tests and FDR corrections were used to control for multiple comparisons.

MVPA identified a reliable neural pattern that accurately classified NU vs. GU conditions (80% accuracy in Sample 1, 85% in Sample 2). This pattern showed increased activity in bilateral DLPFC, DMPFC, left VLPFC, bilateral ACC, bilateral OFC, left AG, left MTG, bilateral thalamus, and right cerebellum; and decreased activity in right superior parietal lobule, right precuneus, and right inferior lateral occipital cortex. Cognitive decoding linked the positive weights to higher-order cognitive processes (emotion, memory, reasoning) and the negative weights to visual processing. The DT pattern showed a high correlation with the primary gradient of functional connectivity, suggesting integration from sensory to abstract cognition. Neurobiological analyses revealed positive correlations with dopamine-related neurotransmitters (MOR, CB1, H3, mGluRs) and genes involved in neurotransmitter release and negative regulation of multicellular organismal processes. RVR demonstrated that the DT brain pattern predicted individual DT scores in both task-based fMRI samples and resting-state functional connectivity data, particularly connectivity between left AG and left MTG.

The study's findings support a distributed neural network model of DT involving the DMN, FPCN, and limbic network for positive weights and visual and sensorimotor networks for negative weights. The strong correlation with the principal gradient of functional connectivity suggests that DT relies on flexible integration across multiple hierarchical levels of processing. The neurobiological findings highlight the role of dopamine and related genes in DT, consistent with the reward and motivational aspects of creative processes. The ability to predict individual DT scores from both task-based and resting-state data suggests that the identified brain pattern is a reliable biomarker for creativity.

This study provides a comprehensive neural, genetic, and cognitive characterization of divergent thinking. The identification of a reliable brain pattern predictive of DT, its connection to specific cognitive processes, and its neurobiological underpinnings advances our understanding of creativity. Future research should investigate other forms of DT and creativity across different domains, utilizing more balanced samples and exploring causal relationships.

The study primarily focused on verbal DT using a modified AUT, which might not fully capture all aspects of creativity. The sample predominantly included female participants, limiting the generalizability of findings. The use of self-reported originality ratings might introduce subjective bias. Future research should address these limitations by examining other forms of DT, using larger and more balanced samples, and incorporating objective measures of creativity.