Test-retest reliability of resting-state EEG intrinsic neural timescales

Abstract

Intrinsic neural timescales (INTs), which reflect the duration of neural information storage within local brain regions and capacity for information integration, are typically measured using autocorrelation windows (ACWs). Extraction of INTs from resting-state brain activity has been extensively applied in psychiatric disease research. Given the potential of INTs as a neural marker for psychiatric disorders, investigating their reliability is crucial. This study, using an open-source database, aimed to evaluate the test-retest reliability of ACW-0 and ACW-50 under both eyes-open and eyes-closed conditions across three sessions. The intraclass correlation coefficients (ICCs) were employed to quantify the reliability of the INTs. Our results showed that INTs exhibited good reliability (ICC > 0.6) at the whole-brain level across different index types and eye states. Spatially, except for the right temporal region in the eyes-open condition, all other regions showed moderate-to-high ICCs. Over 60% of the electrodes demonstrated moderate-to-high INT ICCs under both eyes-open and eyes-closed conditions, with ACW-0 being more stable than ACW-50. This present study comprehensively assessed the reliability of INT under various conditions, providing robust evidence for their stability in neuroscience and psychiatry.