Figure 2. Predictions for RSA.

Schematic depiction of RSA and predictions for a subset of six triads. Triads 1–3 were studied in a blocked manner (pink); triads 7–9 were intermixed (teal). For all matrices, the colour in each cell represents the predicted change in (Δ) RS between an A item (horizontal) and a C item (vertical) from pre- to post-study. Black cells indicate no change in similarity from pre- to post-study; orange cells indicate learning-related increases in similarity; blue cells indicate learning-related decreases in similarity. White cells represent comparisons across learning condition (for example, intermixed item A7 with blocked item C1), which were excluded from all analyses. Matrices depict predicted item similarities for regions showing integration for both blocked and intermixed learning (top left), separation for both blocked and intermixed learning (top right) and blocked → integration (bottom left), and intermixed → integration (bottom right) interactions with learning condition. Inset bar graphs show predicted average similarities across all within- (darker bars) and across-triad (lighter bars) comparisons. Integration and separation were operationalized as significantly greater (integration) or less (separation) within- than across-triad Δ RS.