Similarities and differences in the effects of different stimulus manipulations on accuracy and confidence

Visual stimuli can vary in multiple dimensions that affect accuracy and confidence in a perceptual decision-making task. However, previous studies have typically included just one or at most two manipulations, leaving it unclear whether each manipulation has a unique effect on accuracy vs. confidence. Subjects indicated whether a tilted Gabor patch was oriented clockwise or counterclockwise from 45°. We included manipulations of the task-defining feature (tilt offset) and four auxiliary, non-task-defining features (size, duration, spatial frequency, and noise level). We found that the four auxiliary manipulations had fairly similar effects on accuracy and confidence. In contrast, the task-defining tilt offset manipulation stood out by affecting accuracy more strongly than confidence. In addition, tilt offset exhibited a supraadditive interaction with all other manipulations for both accuracy and confidence, whereas all auxiliary manipulations exhibited either no interactions or subadditive interactions with each other. Furthermore, tilt offset was the only manipulation for which confidence in incorrect trials decreased with increasing difficulty, while all auxiliary manipulations exhibited the opposite trend. Overall, our results reveal a noticeable similarity among the effects of all four auxiliary (non-task-defining) manipulations on accuracy and confidence, as well as a prominent difference between them and the task-defining manipulation (tilt offset). These results enable a priori predictions of how novel manipulations would affect accuracy and confidence.