A product across separate blocks reduced mean gamma score magnitude by over 95 percent for pairs on different shafts in two SEEG rest records

We tested whether a gamma phase relation repeats across separate blocks in two public rest SEEG records. The source was OpenNeuro ds005670 version 1.0.0. The fixed selection supplied 150 pairs on different electrode shafts per record. The band was 70 to 110 Hz, with 98 to 102 Hz removed. Calibration used 240 to 298 seconds. Tests used 300 to 350 seconds and 370 to 420 seconds. The mean squared scores were 0.011532 and 0.010422. The signed block products were -0.000421 and +0.000462. Their network mean magnitudes were 96.35 and 95.57 percent smaller. Conditional block sign tests gave Holm p values of 0.8926 and 0.0703. Neither passed 0.05. A stronger random phase model permitted a small remaining effect in record 02. A follow-up check omitted segments with samples at digital limits. Its score ratio for record 02 rose to 5.64 percent after a calibration change. We proved removal of self-product bias for independent segments on a fixed power scale. The block product estimates repetition if the block means are independent on that scale. The package contains exact selected samples, code, four figures, source hashes, proofs, and a successful full replay. This method test does not establish an absence of brain connections, a stimulation target, or BCI performance.

stereoelectroencephalography · gamma oscillations · phase connectivity · imaginary coherency · sample bias · brain-computer interfaces