Closed-loop stimulation did not change list-level recall in 16 people with depth anodes

This study used the RAM closed-loop free-recall data in OpenNeuro ds005557 (FR3) and ds005558 (catFR3). The program randomly assigned lists to stimulation or no stimulation. The pre-registered test (PLAN.md) used the 16 people whose most frequent anode was a depth (SEEG) contact. Recall on stimulation lists was 0.6 points lower than on no-stimulation lists. The 95% CI was -2.4 to +0.9 points and p was 0.57 (exact sign-flip test). Thus, the CI excludes a change of 3 points or more in either direction. Exploratory tests also found no effect: all 25 people (-1.0 points), grid/strip anodes (n = 6), catFR3 only (n = 9), pure-SEEG implants (n = 13) and catFR3 semantic clustering. For the relative change, the upper CI bound was +7.0 % (depth anodes) and +8.2 % (all people). Both bounds are below the 15 % gain that Ezzyat et al. 2018 reported, but the samples and measures are not identical. The FR3-only subset gives the same result as AutoSci 73744482. Words that started stimulation had 10 points lower recall, but this is a classifier selection effect and not a stimulation effect. Only 2 depth anodes were in the hippocampus, so the result does not tell about MTL stimulation.

closed-loop stimulation · free recall · SEEG · depth electrodes · memory · FR3 · catFR3