Neural Characterization of the Improvisational Creative Process

Mobile Brain-Body Imaging (MoBI) enables the study of the human creative process in freely-behaving participants in natural settings. Past studies on human creativity rely on neuroimaging technology that requires participants to remain in a confined, motionless space. This limits the study design to static, queued actions that oversimplify creative actions. Other studies rely on psychometric tests that compare scores to brain activity at rest, which cannot claim a specific bearing on the creative process.

The main goal of this dissertation is develop novel experimental and analytical approaches to assay the human creative process in natural settings. To accomplish this goal, developed two experiments: 1) We examined the creative process in professional visual artists working collaboratively, in an adaptation of the Exquisite Corpse surrealist game; 2) we examined neural data of college students as they created compositions before and after a 16-week creative writing workshop. These experiments aim to identify and characterize neural features associated with the highly dynamic creative process. We used frequency-domain, time-domain, and functional connectivity features from scalp Electroencephalography (EEG). Both classical machine learning and deep learning approaches were deployed to identify the most relevant features.

Two major findings were obtained. First, the functional connectivity analysis identified patterns between right parietal with left central-frontal scalp areas during creative execution, which were enhanced with experience. Second, the machine learning methods successfully classified neural EEG data in both studies. In the Visual Arts experiment, the classification accuracy reached 53.5 ± 2.4% for 5-classes: two rest conditions, planning, mark making, and writing. In Creative Writing, the classification accuracy reached 79.3 ± 3.1% for 4-classes: two rest conditions, transcription, and creative writing.

Overall, these findings suggest that creative execution tasks can be characterized by a state of long-range cortico-cortical communication between multisensory integration in parietal brain regions and high-order execution and planning areas in frontal regions of the brain. This dissertation provides evidence for common information flow patterns in professional visual artists and student writers matching increased flexibility for creative evocation. In conclusion, this approach provided a better understanding of the human creative process through neural feature characterizations in real world settings.