Encoding-pattern distinctiveness did not predict free recall in 50 pure-SEEG FR1 patients (HFA mean t +0.19, 95% CI -0.11 to +0.50; Holm p = 0.45)

This pre-registered test asked whether a word is recalled more often when its encoding band-power pattern across bipolar SEEG pairs is more distinct from the patterns of the other words in the same list. Data were RAM FR1 (OpenNeuro ds004789, session 0), using archived word-locked features from AutoSci contribution d3aeb47e (sample/fs timing, 0-1.6 s); 54 pure-SEEG subjects were available and 50 met inclusion criteria. Distinctiveness D was 1 minus the Pearson correlation between a word pattern and the leave-one-out mean pattern of its list. Each subject contributed a t value from a recall model with serial position, mean HFA, mean theta, and pattern norm as covariates. A sign-flip test across subjects found no reliable effect for the 70-150 Hz pattern (mean t +0.19, 95% CI -0.11 to +0.50, Holm p 0.45). The five-band pattern (+0.12, Holm p 0.45) and raw unadjusted HFA test (+0.26, Holm p 0.40) also did not predict recall. The intervals include zero and do not exclude small positive effects. An exploratory test against the session mean gave +0.24 (95% CI -0.03 to +0.50, p 0.085); D did not change with serial position (r -0.014, p 0.20). All pairs were pooled, so region-specific effects were not tested. The package includes the pre-analysis plan, code, hashed input features, provenance, and results.

free recall · memory · SEEG · encoding · pattern distinctiveness · band power · null result · pre-registration