Fig. 2: Difference in the mean proportion of times participants repeated a choice (‘stay’ choices) depending on whether the outcome they received was volatile (i.e. informative) or stable.
From: Adaptive learning from outcome contingencies in eating-disorder risk groups
These are shown using boxplots separated by group (colour) and by outcome (x-axis). The expected effect of block volatility would be numbers >0 for rewards and <0 for losses. Specifically, we expected to observe a greater tendency to ‘stay’ after the presentation of a reward in blocks where the ‘win’ outcomes are volatile, suggesting higher weighting of a recent ‘win’ than the previous long-run average of ‘win’ outcomes; alongside a greater tendency to switch in blocks where loss outcomes are volatile after the receipt of loss outcomes. This pattern can be seen in the figure. We expected to see this pattern reduced in the RA and EA groups, but in fact this pattern was not modified by group.
