The Hippocampus is Necessary for the Consolidation of a Task Appendix

Supporting Information for: The hippocampus is necessary for the consolidation of a task that does not require the hippocampus for initial learning Analysis of excluded participants. Four out of eight patients were excluded from the main analyses for failing to meet the inclusion threshold of 10 correctly typed sequences on average across the last three trials of training. Figure S1 shows the learning time courses for the excluded patients, with the included patients for comparison. Three of these patients performed far below threshold (3.7, 3.2, and 3.1 sequences; yellow, purple, and red curves in Fig. S1) whereas one patient was close to the threshold (8.6). We completed the protocol with this latter patient, testing them with a new sequence on a second set of two-day sessions, where performance was again near but below threshold (9.3). Across the two sequences (light blue curve in Fig. S1), this patient had a 29.4% decline in performance for initial change (last three training trials to first three test trials) and an 8.6% improvement for plateau change (last three training trials to last six test trials). These results indicate that this patient showed the same pattern as the included patients: a decrement in initial performance at test without much evidence for improvement by the end of test. We performed an analysis adding this patient as well as the two excluded control subjects (who also performed near criterion, with 8.3 and 9.1 typed sequences) to the participants who passed the inclusion threshold. For initial change, patients exhibited a significant decline in performance (t[4]=3.32, p=0.029, d=1.48), and they showed no difference between the end of training and the test plateau (t[4]=0.05, p=0.964, d=0.02). The difference between patients and controls was marginally significant for initial change (t[15]=1.83, p=0.087, d=0.98) and not reliably different for plateau change (t[15]=1.43, p=0.173, d=0.76). Thus, relaxing the inclusion threshold yielded results very similar to those reported in the main text. The three patients who performed far below the inclusion threshold were performing virtually at floor during training, making it difficult to test for any decrement in performance at test. They showed no evidence of improvement across training: From the first three trials to the last three trials of training, they typed on average 0.2 fewer sequences. For the data collected in these patients (one set of two-day sessions), they exhibited a 13.9% improvement on average in initial change and a 23.8% improvement for plateau change. By the last three trials of test, they were typing an average of only 4 correct sequences per trial. While this is a different pattern than that seen in the other patients, we do not view these results as informative, as having some evidence of memory formed during training is a prerequisite for meaningfully asking how that memory changes at test. Figure S1. Correct sequences across trials of training and test for the four excluded patients, shown with dotted lines, as well as the four included patients (as in Fig. 2b). Curves are averaged across two sequences for the patients that completed two sequences (the top five performing patients).