Predicting personality traits from resting-state fMRI

摘要: Personality neuroscience aims to find associations between brain measures and personality traits. Findings date have been severely limited by a number of factors, including small sample size omission out-of-sample prediction. We capitalized on the recent availability large database, together with emergence specific criteria for best practices in neuroimaging studies individual differences. analyzed resting-state functional magnetic resonance imaging data from 867 young healthy adults Human Connectome Project (HCP) database. attempted predict traits "Big Five", as assessed NEO-FFI test, using connectivity matrices. After regressing out potential confounds such age, sex, IQ, we used cross-validated framework, test-retest replication, quantify how well could each five two superordinate factors ("α" "β"). To obtain more comprehensive set findings, tested three different preprocessing pipelines fMRI data, parcellation schemes, linear models Across all 24 results (test/retest; 3 processing pipelines; 2 schemes; models) found no consistent evidence predictability any exception Openness, β factor ("personal growth/ plasticity"). As benchmark, showed that replicate prior reports predicting IQ same dataset. Best predictions cases were around r=0.2, thus only accounting about 4% variance. conclude discussion make recommendations field.