Expected eight-pulse late ventral temporal SEEG shape energy had a 55 percent mean patient median variance share in five patients

We measured the pulse variance term in squared SEEG response energy. We used OpenNeuro ds004457 version 1.0.2. The study included 590 stimulation block–contact pairs at 19 ventral temporal contacts in five patients with epilepsy. The late window was 100 to 500 ms after stimulation. Each link used 8 to 19 good pulses from one block. For uniform eight-pulse subsets, the mean patient median variance share of expected energy was 54.8 percent. A descriptive patient bootstrap interval was 32.0 to 77.9 percent. The matched pre-window share was 113.9 percent. The late share was 55.9 percent after pulses with source notes were omitted. Values above 100 percent occur because the distinct-pulse estimate can be negative. The finite-pool identities apply to the recorded values. An unbiased signal energy interpretation requires a common mean waveform and zero error covariance between distinct pulses. The variance term includes true response variation as well as measurement error. We checked all 57 required raw files and rebuilt every input array exactly. A second analysis matched all 22 numerical output hashes. The pinned wheel environment also matched every energy value exactly. The package preserves pulse bins, source versions and hashes, link results, pulse membership, code, proofs, and assumption checks. This small cohort does not establish clinical thresholds or false-connection rates.

SEEG · electrical stimulation · pulse variance · squared response energy · finite-pool estimator · ventral temporal cortex