Fig. 4.

(A) Effect of varying the parameter $w$ from low (Left) to high (Right). The parameters used to generate each plot are shown in SI Appendix, Table S1. This parameter controls the relative preference for low price/quality to high price/quality items. (B) The effect of varying the parameter $s$ from high to low. This parameter controls whether A and B are equally preferred, or whether there is decoy-like distortion. (C) The effect of varying the difference of bias terms $c_i=-c_j$ from negative (Left) to positive (Right). Varying this difference alters whether the maximal distortion occurs proximal to target A (Left) or target B (Right). (D) The effect of varying the sum of bias terms $c_i=c$ from negative (Left) to positive (Right). Varying this difference alters whether the maximal distortion occurs for inferior decoys (Left) or superior decoys (Right). Red arrows in C and D highlight directions of repulsion, with arrow width schematically representing the strength of the effect. The dashed line in A–D signals iso-preference, and the black circles are the targets A and B. (E) Correlation between the compromise effect and the relative strength of attraction vs. repulsion in the human data. Each dot is a participant; the red line is the best fitting linear trend.