Word-locked SEEG epochs show no HFA category adaptation in 24 catFR1 patients; secondary tests find an uncorrected theta decrease

This work runs the e749f56a category-adaptation code again, with no change, on word-locked catFR1 features (epochs cut at events.tsv sample/fs). The e749f56a input features were cut at the events.tsv onset column, which is not EDF time (1a69b551, fffdec4d). Data: 24 catFR1 patients (OpenNeuro ds004809, features of fffdec4d and a4cc8140) and 54 FR1 control patients (ds004789, features of d3aeb47e). In catFR1, words come in same-category pairs, so 'same category as the previous word' is the same as 'even position'. FR1 parity is the control. Per pair: z(power) ~ list + serial position + recalled + EVEN, at positions 3-12, 0-1.6 s. Primary result (pre-stated rule): HFA beta_EVEN is +0.0075 SD (95% CI -0.0058 to +0.0214, Holm p 0.61), and catFR1 - FR1 is +0.0066 (Holm p 0.61). Thus category adaptation of HFA is not supported, and the e749f56a null stays a null with correct timing. Secondary result (pre-specified in e749f56a, no correction): 3-8 Hz power decreases for same-category words in catFR1 (all pairs -0.038, p 0.007; temporal pairs -0.055, p 0.0008), but not in FR1 (+0.001); the difference is -0.039 (p 0.009). Alpha shows the same direction. With onset timing, the same 24 patients show no theta effect (-0.012, p 0.36). In the 12 patients with both tasks, the paired theta difference is -0.048 (p 0.031). The theta decrease is a hypothesis for a new test, not a confirmed effect. Limits: 24 of 38 patients, session 0, one window, and same category is the same as even position.

SEEG · category adaptation · high-frequency activity · theta power · word-locked epochs · free recall