A Shared Neural Marker Predicts Creative Performance Across Distinct Problem-Solving Tasks
Creativity is essential for innovation, yet the brain mechanisms supporting its moment-to-moment variability remain unclear. We hypothesize that creativity depends on dynamic fluctuations in neural flexibility, which determine the potential to generate creative solutions. Here, we identify a shared neural marker of 'creativity potential' that predicts upcoming performance across distinct problem-solving tasks. Twenty-eight participants completed the Alternative Uses Test (AUT), a measure of divergent thinking, and the Fusion Innovation Test (FIT), which integrates divergent and convergent thinking. Responses were scored for novelty, feasibility, and goal attainment using validated GPT-based automated evaluation. EEG signals recorded prior to problem onset were used to decode single-trial creativity scores. A decoding model based on coherence features achieved robust performance (r = 0.45, leave-one-trial-out) and generalized across individuals (r = 0.34, leave-one-subject-out). Feature weights revealed a creativity potential network (CPN), characterized by frontal-temporal interactions in beta frequency band. Applying the model to resting-state recordings revealed ~3-minute cycles of creativity potential, suggesting intrinsic brain dynamics shape readiness for creative problem-solving. These findings establish a shared neural marker of creativity that transcends task boundaries and individuals. Beyond advancing our understanding of creative cognition, this work opens the possibility of monitoring creativity potential in real time, with implications for neurofeedback and creativity enhancement in daily life.