A low-variance subspace underlies individual differences in resting state fMRI

Non-invasive whole-brain recordings of human brain activity, such as those from resting state fMRI (rs-fMRI), contain reliable individual differences across thousands of features, but it remains unknown whether this reliability can be isolated in a small number of dimensions. Here, we directly optimize test-retest reliability across many of these features and identify a low-dimensional linear subspace with very high reliability. These dimensions form personal fingerprints, allowing participants to be identified with high accuracy despite fingerprints explaining only a fraction of the total variance. Several dimensions are strongly associated with a single anatomical, demographic, or behavioral variable, and most dimensions can be predicted from the anatomical layout of cortical regions. Together, our findings suggest that stable individual signatures can be isolated from rsfMRI. These signatures reflect persistent anatomical and physiological differences, and provide a principled low-dimensional basis for biomarker discovery.