Figure 6.

Amygdala modulation of MTL activity-memory relationships. Mixed-effects logistic regression results revealed that the level of amygdala activity significantly affected the relationship between (A) PRC activity and subsequent recollection outcomes and (B) hippocampal activity and subsequent source memory outcomes. The formulas for the corresponding regression models are displayed at top (for simplicity, only fixed effects are listed). The 3 panels in each plot depict the probability of subsequent recollection (or source memory) as a function of MTL activity, split according to relative amygdala activity (low amygdala activity = activity more than 1 standard deviation below the mean, high amygdala activity = activity more than 1 standard deviation above the mean). Plots depict the best fitting logistic regression model relating binary memory outcomes to activity levels, such that the significant interactions are reflected as the change in regression slopes across different levels of amygdala activity. The top row of plots show the fits for emotional trials (pink lines) and the bottom row of plots show the fits for neutral trials (green lines). Interaction effects did not significantly differ between emotional and neutral for either analysis. Rug plots (gray ticks) show the distributions of all trials across activity levels, divided into the corresponding memory outcomes (top rug = subsequently recollected or R+S, bottom rug = non-recollected or R-S). For the purposes of visualization, trials were aggregated across subjects, but the regression models and resulting slopes were estimated with subject and session as nested random effects.