Figure 3.

Comparison of Subjective Distributions Across Tasks and Subjects

(A) Jensen-Shannon (JS) distances between subjective distributions inferred in the same subject for the two different tasks (left), inferred in different subjects within the familiarity (middle) and odd-one-out (right) tasks. Dots show individual comparisons (left, subjects; middle and right, subject pairs), boxes show mean ± SE. The dashed line shows the estimated lower bound based on the average distance between distributions inferred from two halves of the data from the same task and same subject. ^∗p < 0.05, ^∗∗p < 0.01.

(B) Two-dimensional embedding of subjective distributions for the ten subjects and two tasks (symbols) based on multidimensional scaling applied to all 190 pairwise JS distances. Lines connect distributions of the same subject, and line width is proportional to the consistency score of the subject.

(C) Across-task JS distances for each subject (symbols) against the subject’s task-average consistency score. The regression line shows hyperbolic fit to data.

Colors for subjects in (B) and (C) are as in Figure 2. See also Figure S3.