In RAM FR2 depth-anode data, encoding stimulation did not measurably delay recalled words

This pre-planned analysis tested whether the recall decrease associated with open-loop 50 Hz stimulation during word encoding was all-or-none or graded among words that were still recalled. We compared stimulated and non-stimulated words from the same lists, with list and serial-position fixed effects. In the primary sample (FR2, depth anodes, 26 subjects for output position), stimulation was not associated with a later output position (+0.005; 95% CI -0.049 to 0.055; two-sided sign-flip p=0.86), slower log latency (+0.026; CI -0.070 to 0.121; p=0.60), or lower probability of first recall (-0.037; CI -0.107 to 0.024; p=0.30). The output-position confidence interval excludes shifts larger than about 5% of the recall sequence. Thus, the primary test did not support a graded effect; all-or-none was not rejected. In catFR2 depth-anode data (n=9), output position showed a small trend in the graded direction (+0.082; CI 0.016 to 0.148; p=0.051), which was not significant. Within-list recall decreases were small, so power to detect a graded effect was low. The analysis did not model temporal contiguity, category clustering, electrode location, or stimulation amplitude.

free recall · memory · open-loop stimulation · 50 Hz · FR2 · catFR2 · SEEG · recall order · latency · OpenNeuro