(A) Choice behavior for the one reference per block and two references per block conditions for all participants in Experiment 2. (B) Decisional and sensory biases for each participant. The color indicates the best model. (C) Across the participants of Experiment 2, a two-way repeated measures ANOVA for the absolute value of the biases indicated no significant effect of number of references per block (one reference per block or two references per block; (F(1,14) = 3.7; p=0.073), a significant effect of the type of biases (sensory or decisional; F(1,14) = 8.7; p=0.010) and no significant interaction (F(1,14) = 3.6; p=0.078). The graph shows the biases from Experiment 1 (also shown in Figure 1H) and Experiment 2 (collapsed across number of references per block). The crossbars display the mean and the 95% t-test confidence intervals. For Experiment 2, the sensory and the decisional biases is significantly different (t(29) = 3.13, p=0.0040). Across experiments, the sensory biases is not significantly different (t(47) = 0.92, p=0.3619), but the decisional biases is significantly different (t(44) = 2.0, p=0.048).