Fig. 5

The effect of similarity on accuracy in the post-experiment tests—the recall tests (average of sessions #1–3 in each week) and the two-alternative forced-choice tests (session #4 in each week). Only the trained facts are shown here. The p values above the columns are the effect of the Numeric Similarity factor in the LLMM described in the text. The p values inside the forced-choice columns refer to comparing the per-participant mean accuracies to chance level (paired t-test against 50%, one-tailed p)