Sampling Rate and Task Selection in EEG-Authentication

EEG-based authentication is a promising security alternative, measuring unique neural patterns, but variability in sampling rate (fs) across datasets can distort performance metrics. This study evaluates how fs normalisation impacts authentication accuracy by analysing multiple EEG datasets with differing fs. Using machine learning classification pipeline, we show that optimal fs selection depends critically on task-specific neural dynamics: high-frequency gamma tasks require≥ 500 Hz, while alpha-dominated paradigms perform well at 128 Hz, and auditory potentials remain stable even at 98 Hz. Notably, prefrontal tasks show inherent limitations unaffected by fs. Findings emphasise the need for standardised paradigm-specific fs in EEG authentication to improve reproducibility and robustness. This work provides practical insights for optimising biometric systems and advancing EEG-based authentication. Code available upon request.