In RAM FR2 events from 24–30 patients, preceding-word recall did not identify state-dependent effects of open-loop 50 Hz encoding stimulation

We tested whether the effect of open-loop 50 Hz stimulation during word encoding depended on a behavioral marker of encoding state: recall of the preceding, non-stimulated word. We analyzed only FR2 event files from OpenNeuro ds005489 (78 sessions, 38 subjects); the electrophysiology signals were not used. Targets were serial positions 5 and 9 in stimulated lists. For each patient, we estimated stimulation effects after a preceding word that was not recalled (NR) and after one that was recalled (R), using two controls: positions 5/9 in no-stimulation lists (n=24) and non-stimulated positions 4/8/12 in stimulation lists (n=30). The marker predicted recall of the next word in control words (+0.122, 95% CI 0.029 to 0.218, p=0.023; +0.257, 0.173 to 0.345, p<0.0001) and in stimulated words (+0.171, 0.099 to 0.244, p=0.0001; +0.176, 0.109 to 0.244, p<0.0001). The state-by-stimulation interactions were −0.049 (95% CI −0.146 to 0.048, p=0.34) and +0.081 (−0.023 to 0.195, p=0.17). The estimates had opposite signs and both confidence intervals included zero. Thus, this behavioral marker did not identify words for which open-loop stimulation helped or harmed recall. The test cannot exclude interactions of about ±0.15 or a dependence on neural state. Temporal contiguity may affect the marker, especially for the secondary control. Other limitations include few no-stimulation lists, pooled stimulation sites and amplitudes, uncorrected p-values across 12 tests, and targets at positions 5 and 9 only. The package includes standard-library Python code, source event files, SHA-256 provenance, per-patient tables, and a figure.

memory · free recall · encoding stimulation · closed-loop stimulation · behavioral state · temporal contiguity · RAM FR2 · intracranial EEG