Word frequency and length predict recall but do not change mean encoding HFA in English ds004789 FR1 sessions

We tested whether word frequency (Zipf) and word length changed mean encoding high-frequency activity (HFA, 70–150 Hz) or theta power (3–8 Hz) in ventral temporal cortex (VTC) and hippocampus (HIP) during delayed free recall in OpenNeuro ds004789 FR1, session 0. Neural analyses used archived per-word features and event files; frequency came from the OpenSubtitles 2018 English list. In the English sessions, VTC HFA effects were near zero for length (n = 46, beta = −0.0020 z per SD, 95% CI −0.0121 to +0.0082, p = 0.71) and frequency (n = 46, beta = −0.0075, 95% CI −0.0198 to +0.0040, Holm p = 0.45). HIP HFA, VTC and HIP theta, and left and right VTC analyses were also null. All neural 95% confidence intervals were within ±0.021 z per SD. In recall models covering 237 subjects, longer words were recalled less (beta = −0.0095 probability per SD, p < 10⁻⁴), while frequency showed a U-shaped association (linear beta = +0.0060, p = 0.013; squared-frequency beta = +0.0104, p < 10⁻⁴). Six sessions used Spanish word pools and were removed after the first run because English frequency coverage was below 95%; the first-run analyses gave the same conclusions. The positive control was weak, the single 0–1.6 s window could miss early lexical effects, and the word pool had a narrow range. Thus small effects may have been missed; the result is limited to the reported measures, sessions, and population.

free recall · word frequency · word length · high-frequency activity · ventral temporal cortex · hippocampus · stereo-EEG · OpenNeuro ds004789