Theta–gamma coupling was lower for recalled words, with no power-independent effect detected in 35 SEEG patients

We tested a pre-registered question: whether hippocampal theta–gamma phase–amplitude coupling (PAC) during word encoding was larger for words recalled later. We analyzed session 0 of the RAM FR1 SEEG dataset (OpenNeuro ds004789), using hippocampal bipolar channels from 35 subjects, 292 pairs, and 8,616 word epochs. The Tort modulation index was computed with trial counts matched between recalled and not-recalled words; subjects were the unit of inference. Coupling from 3–8 Hz phase to 30–55 Hz amplitude was lower for recalled words (mean log-MI difference −0.162, 95% CI [−0.275, −0.054], sign-flip p = 0.0068, Holm p = 0.014). The difference for 70–150 Hz amplitude was not significant (−0.094, 95% CI [−0.193, 0.005], Holm p = 0.080). Theta power was also lower for recalled words (p = 0.0045). In a per-epoch model that included theta and gamma power, the recall effect on MI was near zero (30–55 Hz: −0.003 SD, p = 0.54). Exploratory analyses found that the PAC decrease tracked the theta-power decrease (r = 0.40, permutation p = 0.016). The MI decrease may therefore result from lower theta power and a less clear phase estimate. These data do not show a recall-related coupling change independent of power. The study covers one task and session and 35 of 100 listed subjects; its per-epoch control is noisier than its primary pooled measure, so its null result does not prove that power explains all of the primary effect. Interictal discharges were not removed.

hippocampus · theta–gamma coupling · phase–amplitude coupling · free recall · SEEG · subsequent memory · power control