Fig. 3: Model correlation distribution for each emotion category.

Distribution of bootstrapped Pearson correlation of the models with the empirical judgments from joint-cue condition for emotion categories derived from n = 604 pairs of observations. Data is presented where the correlation of the Bayes cue-integration model with participant judgments was a significantly higher than that of the Situation-only model; b not significantly different from the Situation-only model and; c significantly lower than the Situation-only model. The center line in each boxplot represents the mean correlation for the bootstrapped samples. The upper and lower limits of each boxplot represent the upper (75th percentile) and lower (25th percentile) quartiles of correlation values for each model and the whiskers extending from the box represent the 1.5x interquartile range. See Supplementary Table S3 for values of 95% confidence intervals for difference in correlation between pairwise comparison of the three models for each emotion category.