Figure 6.

Effects of self-consistency on confidence judgments. (A) The probability to report higher confidence in the orientation than the color decision is plotted as a function of the difference in sensory distances to the subjective criterion (abscissa) and to the objective criterion (ordinates) for the same observer as Figure 2. Sensory distance to the subjective criterion reflects self-consistency, and sensory distance to the objective criterion reflects correctness. (B) Same as (A) averaged across observers. (C, D) Same as (A) and (B) for the prior probability experiment. (E) Two predictions of probit models where observers rely on the subjective sensory distance only (top) or objective sensory distance only (bottom). Regression weights associated with these two examples are plotted for reference in (G), where the purple and green circles represent the subjective and objective models, respectively. (F–G) Probit regression weights of confidence judgments. Observers’ confidence judgments reflect self-consistency (sensory distance to the subjective criterion), not correctness (sensory distance to the objective criterion).