In Cortex; a journal devoted to the study of the nervous system and behavior
Extensive neuroimaging research has attempted to identify neural correlates and predictors of decision impulsivity. However, the nature and extent of decision impulsivity-brain association have varied substantially across studies, likely due to small sample sizes, limited image quality, different imaging measurement selections, and non-specific methodologies. The objective of this study was to develop a reliable predictive model of decision impulsivity-brain relationship in a large sample by applying connectome-based predictive modeling (CPM), a recently developed machine learning approach, to whole-brain functional connectivity data ("neural fingerprints"). For 809 healthy young participants from the Human Connectome Project, high-quality resting-state functional MRI data were utilized to construct brain functional connectome and delay discounting test was used to assess decision impulsivity. Then, CPM with leave-one-out cross-validation was conducted to predict individual decision impulsivity from whole-brain functional connectivity. We found that CPM successfully and reliably predicted the delay discounting scores in novel individuals. Moreover, different feature selection thresholds, parcellation strategies and cross-validation approaches did not significantly influence the prediction results. At the neural level, we observed that the decision impulsivity-associated functional networks included brain regions within default-mode, subcortical, somato-motor, dorsal attention, and visual systems, suggesting that decision impulsivity emerges from highly integrated connections involving multiple intrinsic networks. Our findings not only may expand existing knowledge regarding the neural mechanism of decision impulsivity, but also may present a workable route towards translation of brain imaging findings into real-world economic decision-making.
Cai Huanhuan, Chen Jingyao, Liu Siyu, Zhu Jiajia, Yu Yongqiang
Decision impulsivity, Delay discounting, Functional connectivity, Machine learning, Predictive model, Resting-state fMRI