Task and design of the experiments.

(A) Time course of one trial. Following a fixation cross, subjects saw 70 red vertical lines sequentially presented on a white horizontal axis. Their task was to move a blue vertical line to indicate their estimate of the Mean (the average horizontal position of the red lines) and then a blue box to indicate the Mode (the area that would catch the largest number of red lines). (B) An example of a 3-beta mix (i.e. mixture of three beta distributions) underlying distribution used in Experiments 1, 3 and S1. This example consists of three negatively skewed beta components (“LocalNeg”), with the weights of the three components from left to right being 0.2, 0.3 and 0.5 (“weights = 235”). (C) An example of a 4-beta mix (i.e. mixture of four beta distributions) underlying distribution used in Experiments 2 and 3. This example consists of four positively skewed beta components (“LocalPos”), with the weights of the four components being 0.1, 0.2, 0.4 and 0.3 (“weights = 1243”). (D) All 3-beta mix underlying distributions used in Experiments 1, 3 and S1. “Pos”, “Sym”, and “Neg” are abbreviations respectively for positive, symmetric, and negative. Local skewness (LocalPos, LocalSym, and LocalNeg) refers to the skewness of the beta components and is controlled by the shape condition. Global skewness (GlobalPos, GlobalSym, and GlobalNeg) refers to the skewness of the whole distribution and is controlled by the weight condition.