Fig. 2.

Neural overlap correlates with canonical false-memory likelihood in the left TP. (A) A whole-brain searchlight analysis revealed a significant cluster in the left TP (peak MNI coordinates: −51, 17, −25; T = 5.33; cluster extent = 92 voxels), with no other region displaying any significant information. Results are displayed on a cortical surface map using BrainNet Viewer (50). (B) To visualize the relationship between neural overlap and false memory, we plot the group average neural overlap for each of the 40 DRM lists against canonical false-memory likelihood, using a cross-validation procedure over subjects to avoid artificial inflation of the effect size. There is a clear positive correlation between the two [r(39) = 0.40, P = 0.012], showing that the degree of semantic relatedness in the neural data predicts variation in false-memory strength across the DRM lists. This correlation remains [r(39) = 0.45, P = 0.005] after removing two potential outliers.