Figure 3.

Stimulus presence downsampling analysis. A, For each participant, trials were deleted until confidence distributions were matched for target present and target absent responses. As a control analysis, we repeated this procedure with random downsampling, deleting the same number of trials regardless of confidence ratings. Error bars indicate the standard error of the mean. B, Presence/absence classification accuracy in the four ROIs for the equal confidence and random downsampling datasets. *p < 0.5. ***p < 0.001.