In two subjects, SEEG contacts had a lower fitted DN exponent than same-region ECoG contacts

We compared published delayed divisive-normalization (DN) fits for broadband (70–200 Hz) visual-cortex responses from SEEG depth and ECoG surface contacts. In four subject-by-region strata from two subjects, 19 depth and 9 surface contacts were compared. The fitted exponent n was lower for depth contacts (depth minus surface = −0.225; 95% bootstrap CI −0.300 to −0.148; permutation p = 0.0039; Bonferroni p = 0.0195). Cross-validated r², tau, tau_a, sigma, time to peak, and sustained-to-peak ratio showed no detected difference. The plan's alpha = 0.01 per outcome threshold is met by the unadjusted p value, while its p < 0.01 after correction criterion is not met. Only two subjects and nine surface contacts contributed; contacts on one lead are not independent, three strata had one surface contact, and n trades off with sigma. The lower fitted n therefore does not establish a difference in neural nonlinearity. The source files came from an untagged OpenNeuro S3 state, and their exact dataset release is uncertain. The model was not refit.

visual cortex · SEEG · ECoG · divisive normalization · electrophysiology · temporal dynamics