A pre-registered catFR1 replication found no significant recall decrease after hippocampal IED detections

We tested whether hippocampal interictal epileptiform discharge (IED) detections during word encoding predict lower later recall in an independent cohort of categorized free-recall (catFR1) patients. A plan was hashed before signal or behavior was fetched. We applied the same detector and analysis code as a prior FR1 result to the 54 catFR1 patients with hippocampal depth contacts who were absent from FR1 dataset ds004789. At the preregistered threshold, 26 patients met the inclusion rule. Their mean recall difference for words with versus without a detection in the 0–1.6 s encoding window was −2.1 percentage points (one-sided sign-flip p = 0.115; Holm p = 0.13; bootstrap 95% CI −5.5 to +1.1; 14 of 26 patients had negative differences). The serial-position-stratified result was −2.7 points (Holm p = 0.13). By the preregistered rule, this did not replicate the prior −4.3-point FR1 result. The confidence interval includes −4.3 points, so the data do not reject an effect of that size; only 26 of 54 patients met the detection-based inclusion rule. A prespecified pre-stimulus control window gave a similar decrease (−1.7 points), and a post hoc paired encoding-minus-control difference was −1.2 points (two-sided p = 0.65), so the data do not show an encoding-specific effect. Exploratory threshold and hemisphere results were uncorrected and are not findings. The automatic detector was not checked against expert marks in this cohort, and words were not matched for list, session time, or category. The package preserves the plan, code, results, source URLs, and byte-level provenance; the exact dataset release used cannot be established from the recorded live-mirror URLs.

hippocampus · interictal epileptiform discharges · free recall · word encoding · replication · intracranial EEG · catFR1