Surface CCEP N1 latency decreased with age across SNR thresholds, but the association did not survive Holm correction

This robustness study tested whether the reported decrease in surface electrocorticography (ECoG) CCEP N1 latency with age depends on the signal-to-noise ratio (SNR) threshold. It used archived per-unit rows from 16 patients (age 5–50; 20,186 N1 units at distances of at least 10 mm), a fixed 2.6 SD N1 detector, and the published distance adjustment. Across 12 thresholds (2.6–30 SD), surface Spearman correlations were negative (rho −0.39 to −0.62). The 8 SD result had rho −0.62 and pointwise p = 0.012; family-wise p was 0.039 over the surface threshold grid and 0.14 over both surface and depth contact types. Holm correction of the 18 published tests gave a minimum adjusted p = 0.225. Leave-one-patient-out correlations remained negative, and the 8 SD patient-bootstrap 95% CI was −0.88 to −0.09. Depth contacts showed no negative age relation; at 8 SD the estimate was positive (rho +0.49, 14 patients), but prior work indicates that most depth detections below 15 SD may not be true N1. Thus the direction of the surface association was consistent across thresholds and patients, while its strength depended on the test family. The small, single-center cohort, fixed detector, Euclidean distance, confounding of age with electrode plan and pathology, and unregistered grid and 30 mm reference limit inference. The result does not establish an age covariate for depth-contact timing models.

cortico-cortical evoked potentials · N1 latency · age · electrocorticography · stereoelectroencephalography · threshold robustness · permutation test · brain-computer interface