In 38 SEEG patients, encoding HFA was not reliably higher before category-switch than within-category recall (+0.032 z; 95% CI -0.011 to +0.081)

Semantic-search models say that a recall within a category cluster gets help from the previous recall. A recall after a category switch does not get this help. Thus, words later recalled after a switch may have stronger item encoding. We tested this prediction with encoding high-frequency activity (HFA, 70-150 Hz). Data: the archived per-word features of AutoSci 63055d12 (OpenNeuro ds004809, RAM catFR1, session 0, 38 pure-SEEG subjects, bipolar pairs, 0-1.6 s after word onset) and the events.tsv files (SHA-256 checked, 38 of 38 agree). A pre-registered plan was written before features were joined to labels. Each recall at output position 2 or later was labeled switch or within by the category of the previous correct recall (899 switch, 1277 within). Per subject, we used list-centred z-scored HFA averaged over pairs and an OLS model with serial-position covariates. We used a sign-flip test over subjects. Primary result: switch minus within was +0.032 z (95% CI -0.011 to +0.081, 22 of 38 positive, p = 0.19). Thus the hypothesis was not supported. Theta (3-8 Hz) did not differ (-0.006 z, Holm p = 0.79). Behaviour confirmed the labels: 58.7% of transitions were within a category (shuffle chance 44.3%, 35 of 38 subjects), and switch transitions were 1.82 s slower (37 of 38). Limits: the CI upper limit is larger than the whole HFA subsequent memory effect in these features, so a small effect is not excluded. The features average all SEEG pairs, including white matter. Output position is not controlled. Assumptions: the archived features measure encoding HFA correctly, and the events.tsv category labels are correct.

categorized free recall · category clustering · encoding · high-frequency activity · intracranial EEG · semantic memory · SEEG