Periodic Reporting for period 1 - CreativeBrain (The neural mechanism of scale-invariant creative search)
Berichtszeitraum: 2023-05-01 bis 2025-10-31
The human brain efficiently searches enormous mental spaces of thought for solutions to a given question. How is this done? The growing importance of creative search has sparked a surge of interest in mapping the relations between people’s creative search strategies and the structure, activity, and connectivity of their brain. Yet to date there are no coherent computational principles to bind behavioral, computational, and neurobiological findings together into a mechanistic understanding of creative search. CreativeBrain uses two such computational principles: scale-invariant sensing and Pareto optimality. Scale-invariant sensing is essential for a robust search in environments with signals that span many orders of magnitude like in creative search. Pareto optimality asserts that individual differences stem from different balances between competing tasks that need to be optimized and thus explains the utility of these individual differences. CreativeBrain will employ state-of-the-art computational and analysis methods from systems biology to infer the neural mechanism of creative search on the levels of function, computations and neural implementation. This will be the first comprehensive research effort that ties these findings to a mechanistic theory of creative search that also explains the utility of neural individual differences. The project will result in a breakthrough in our understanding of how the human brain can efficiently search in vast spaces of thought. On a broader scale, CreativeBrain will open a new front in the interdisciplinary studies of the mind and brain, offering a principled way to unite neurobiological, behavioral, and computational aspects into one holistic and mechanistic view. By doing so it will contribute significantly to the promise of computational modeling for connecting different levels of inquiry of a higher cognitive function in the brain and can thus be extended to other cognitive processes and systems.
We developed and incorporated the following pipelines and analysis methods towards the analysis of qMRI, fMRI and behavior data:
- A state-of-the-art qMRI protocol for obtaining high-resolution and highly efficient quantitative MR maps.
- We use a state-of-the-art multi-echo protocol resulting in high signal-to-noise ratio (SNR) from both cortical sub-cortical regions.
- We integrated a pre-processing fMRI pipeline that combines different processing protocols and improved their denoising process
- We integrated several cutting-edge structural MRI data pipelines to quantify microstructural features along fiber tracts
- We developed a novel Pareto analysis pipeline for functional-connectivity data during rest and task and similarly for structural connectivity data
We developed new computational tools and models of the creative search process in the Creative Foraging Game (CFG):
- We developed a new analysis methodology for assessing the creative search process during the CFG that provides new embeddings space for shapes in the CFG (Manuscript in preparation).
- We analyzed the exploration in creative process to show that it is not random but driven by a power-law process driven by a scalar symmetry field (Manuscript in preparation).
- We developed methodology to extract neural activity and connectivity pattern from explore and exploit phases during participants' play in the CFG. We then developed a novel analysis pipeline for predicting explore-exploit states from the activity and functional connectivity between large brain networks.
- We are developing a tailored method to evaluate neural dynamics based on the general properties of our updated computational model.
We developed and analyzed the neural infrastructure of creative search:
- We identify resting-state connectivity and structural measures of creative search implementation and predict creative search behavior (Manuscript in preparation).
- A state-of-the-art qMRI protocol for obtaining high-resolution and highly efficient quantitative MR maps.
- We use a state-of-the-art multi-echo protocol resulting in high signal-to-noise ratio (SNR) from both cortical sub-cortical regions.
- We integrated a pre-processing fMRI pipeline that combines different processing protocols and improved their denoising process
- We integrated several cutting-edge structural MRI data pipelines to quantify microstructural features along fiber tracts
- We developed a novel Pareto analysis pipeline for functional-connectivity data during rest and task and similarly for structural connectivity data
We developed new computational tools and models of the creative search process in the Creative Foraging Game (CFG):
- We developed a new analysis methodology for assessing the creative search process during the CFG that provides new embeddings space for shapes in the CFG (Manuscript in preparation).
- We analyzed the exploration in creative process to show that it is not random but driven by a power-law process driven by a scalar symmetry field (Manuscript in preparation).
- We developed methodology to extract neural activity and connectivity pattern from explore and exploit phases during participants' play in the CFG. We then developed a novel analysis pipeline for predicting explore-exploit states from the activity and functional connectivity between large brain networks.
- We are developing a tailored method to evaluate neural dynamics based on the general properties of our updated computational model.
We developed and analyzed the neural infrastructure of creative search:
- We identify resting-state connectivity and structural measures of creative search implementation and predict creative search behavior (Manuscript in preparation).
Our novel application of Pareto analysis to extract diverse creative search computational strategies significantly goes beyond state-of-the-art methodologies that link computational goals with brain function or structure in creativity. This approach holds the potential to catalyze a paradigm shift in computational neuroscience. We are approaching this objective through the development of analytic pipelines that uniquely extract and integrate brain functional connectivity, activity patterns, and structural features, along with our expanded computational model of scale-invariant creative search. The full infrastructure for Pareto-based analysis is now in place and ready for deployment upon reaching the required sample size for rigorous evaluation.
On a more focused scale, our unique embeddings method, applied to task-specific data, introduces a unified framework for investigating creative search across different tasks and domains. This approach can potentially extend to the study of broader cognitive search processes, offering finer resolution and enhanced rigor in characterizing mental search processes in creativity, memory, and decision-making. The embeddings method emerged unexpectedly from the need to improve the accuracy of our explore-exploit segmentation algorithm during participants’ scanning sessions in the magnet. While in the scanner, participants interact using a specialized mouse under motion constraints, which alters the temporal dynamics of their task performance. The prior segmentation algorithm exhibited some inconsistencies in accurately distinguishing between the two phases. The new method, augmented with information from the embeddings space, substantially enhances the precision and rigor of explore-exploit phase detection, thereby strengthening the robustness and validity of our findings.
With regard to creative exploration, our findings reconceptualize it as a directed search on a pruned manifold based on symmetry assessment. This framework introduces a new computational mechanism for understanding the balance of novelty and appropriateness in creative cognition.
On a more focused scale, our unique embeddings method, applied to task-specific data, introduces a unified framework for investigating creative search across different tasks and domains. This approach can potentially extend to the study of broader cognitive search processes, offering finer resolution and enhanced rigor in characterizing mental search processes in creativity, memory, and decision-making. The embeddings method emerged unexpectedly from the need to improve the accuracy of our explore-exploit segmentation algorithm during participants’ scanning sessions in the magnet. While in the scanner, participants interact using a specialized mouse under motion constraints, which alters the temporal dynamics of their task performance. The prior segmentation algorithm exhibited some inconsistencies in accurately distinguishing between the two phases. The new method, augmented with information from the embeddings space, substantially enhances the precision and rigor of explore-exploit phase detection, thereby strengthening the robustness and validity of our findings.
With regard to creative exploration, our findings reconceptualize it as a directed search on a pruned manifold based on symmetry assessment. This framework introduces a new computational mechanism for understanding the balance of novelty and appropriateness in creative cognition.